Conventional web sites versus content management systems

Description

By a conventional web site I mean a filestore directory that a web server is configured to serve as the root of the web site or sub-site together with a convention about the file that is served by default. For example, my personal academic web site for the School of Mathematical Sciences (SMS) at Queen Mary is a conventional web site with root directory "~fjw/public_html" and by default the file "index.html" is served. This directory is on a Unix-like network. In future, I will use Unix to mean Unix-like. For staff who use Unix on their office computers, edit access to their web site is very easy because their web root directory appears directly in their file system. Staff can very easily put any files they want onto their web site using whatever tools they wish. There are no constraints and if they make a mistake their content will not appear correctly on the web.

By a content management system (CMS) I mean a system such as Drupal or QMplus (Moodle/Mahara), which does not provide direct edit access to files to be served. All operations on the web site are managed through a web interface, which supports creation, deletion, modification and general management of all content. A CMS controls to a greater or lesser extent the file types and the content that are allowed on the web site. Typically, one cannot edit the whole of a page that might be served, but only portions of it, and the page that is served is constructed from a number of components, either dynamically or during (say) an overnight rebuild. The general look and feel of the web site is determined by the designers of the site and it is difficult or impossible for a normal user to do much to change this.

A CMS may support version control, by which I mean that it stores every version of all content that is saved and provides the ability to revert to a previous version. Drupal provides this facility whereas QMplus does not (as far as I am aware). A CMS may also offer the facility to develop content offline and only publish it to the web server when it is ready to be published. Drupal offers this facility to some extent by offering a preview of edited content before it is saved whereas QMplus does not. This means that with QMplus the only way to see what edited content really looks like is to save it, at which point all users also see the edited content complete with any errors. This is, in my view, a major failing of QMplus. In an ideal world, the graphical view of web content provided by an editor would show exactly what it will look like on the web, but my experience to date has been that this is not the case. The rendering in many graphical HTML editors, especially those hosted by web browsers, is incapable of showing more than an indication of gross styling and the content can look completely different when viewed normally. I presume the problem is that online HTML editors need to use stylesheets that support the editing process and so cannot simultaneously support the stylesheets that will be used to display the content normally.

There is an intermediate situation between a conventional web site and a CMS, as exemplified by Adobe Dreamweaver (or Microsoft FrontPage, which is now obsolete). A Dreamweaver web site is based on the use of Dreamweaver templates, which are HTML files containing HTML content and HTML comments that indicate to Dreamweaver which sections of the file can be edited and which cannot. Provided all users use Dreamweaver and obey its conventions, everything works as intended. But there is nothing to stop a user from using a different editor (or, in fact, from using Dreamweaver but ignoring the conventions). If this happens, the web site can rapidly unravel. The main SMS web site was at one time a Dreamweaver site, but as it evolved and more academic staff began to edit it, not all those who edited it used Dreamweaver, so the structure began to unravel. Part of the problem was that most SMS academic staff use Unix and Dreamweaver was not available. This was the main reason why it was decided to use a CMS for the main SMS web site, for which Drupal was chosen. A CMS using web-based clients for its management should be independent of the client platform, which I believe is largely true for both Drupal and QMplus.

A more common (and perhaps better) way to maintain a conventional web site than by direct access to the filestore is to use ftp (or something similar, such as sftp or webdav). SMS personal web sites support sftp, which makes it easy to edit remotely and means that even local users do not need to have the web site filestore mounted on their local machine, which may not be easy or even possible for non-Unix users. A CMS will probably not support this mode of access, although Blackboard (which Queen Mary used as its VLE before QMplus) supported webdav and so did Dreamweaver, so I frequently used Dreamweaver to edit my Blackboard site offline. I'm not aware that this is possible with either Drupal or QMplus, which I think is at least partly because they store content in a database rather than as distinct files. One advantage of use ftp or equivalent to maintain a web site is that one can develop content locally and test it fairly thoroughly (possibly using a local web server, as provided by many web IDEs, such as Microsoft Visual Studio – I use Visual Web Developer Express) before publishing it to the main web server, which as I remarked above is a facility that QMplus is lacking.

Evidence

I have developed and/or maintained a number of web sites, not all of which still exist, some of which I will describe below.

Conventional web sites

When the REDUCE computer algebra system became Open Source in 2008, I set up the web site for it at reduce-algebra.sourceforge.net. I based it on the placeholder site provided by SourceForge and the reduce-algebra.com web site set up by the originator of REDUCE, Tony Hearn. I designed this page, wrote or edited the content, produced the image of a REDUCE session and maintain the site (sporadically). SourceForge provides sftp access to the web sites it hosts.

When I started running a one-week residential course for A-level Maths teachers sponsored by the Goldsmiths' Company, I set up a web site at www.maths.qmul.ac.uk/~fjw/goldsmiths to support the course and provide access to teaching materials. I also arranged for the more generic URL www.maths.qmul.ac.uk/goldsmiths to redirect to the real web site. This web site started life as a Dreamweaver site, which I created by copying and modifying the template used for the main SMS web site. But at some point after our main web site moved to Drupal, the template structure to which I was linking was deleted, so I had to rebuild the web site without using a template. There are still some relics of Dreamweaver lurking in the source code. The files for this web site are in my personal file space on the SMS network, which I can access in various ways: it is mounted on my Windows computer via Samba but I often use sftp, especially if I am working from home.

My official SMS home web page URL is www.maths.qmul.ac.uk/~fjw, which is a conventional web site, but I have not really maintained a home page at that URL for a very long time. When I ran a web server on my office computer, centaur, I set up www.maths.qmul.ac.uk/~fjw to redirect to centaur.maths.qmul.ac.uk, When I stopped running my own web server, I needed to create a new home page somewhere, so I decided to make use of our then new Mahara site, now called QMplus Hub. Hence, I set up www.maths.qmul.ac.uk/~fjw to redirect to hub.qmplus.qmul.ac.uk/user/view.php?id=154. (SMS personal web sites run on an Apache server that supports some local configuration, including redirection, via ".htaccess" files.) Nevertheless, I still provide some conventional sub-sites that are directly accessible, such www.maths.qmul.ac.uk/~fjw/modules-2011-12, www.maths.qmul.ac.uk/~fjw/modules-2012-13 and www.maths.qmul.ac.uk/~fjw/modules-2013-14. These are conventional web sites, but the main content is generated from an XML master file, to which I return below.

Microsoft FrontPage and Adobe Dreamweaver sites

These no longer exist (except for the old SMS staff intranet site, which is a Dreamweaver site but it is not publicly accessible). However, the Internet Archive Wayback Machine provides examples of two such sites that I maintained.

The web site that was served by my office computer at http://centaur.maths.qmul.ac.uk was saved by the Wayback Machine 46 times between August 17, 2001 and January 12, 2012. Looking at the first and last copies shows an interesting progression from what may not have been initially a FrontPage site to one that definitely was. In fact, the first copy shows one of the main reasons why I ran my own web server, which was to host interactive demonstrations of differential equation solvers as part of the CATHODE European collaborative research project. For this demonstration I used CGI (Common Gateway Interface) with a Perl program interfacing between the web page and REDUCE. I was not allowed to use CGI on the SMS web server (although, amusingly, I was later allowed to run PHP on the SMS web server, not to run REDUCE demonstrations but to construct navigation automatically for our undergraduate website, which made maintenance very much easier).

The SMS website at http://www.maths.qmul.ac.uk was saved 193 times between July 25, 2001 and February 22, 2016. I believe that I maintained the undergraduate section from around 2007 onwards. For example, several links are to centaur. The source code shows several HTML comments of the form

<!-- InstanceBeginEditable name="breadcrumb" -->

which are from the Dreamweaver template. The snapshot on 30 June 2009 of the Undergraduate section shows some of the changes I had introduced, including a set of images of interesting two-dimensional surfaces in the right-hand column that I generated using Maple. (I still use one of them on my QMplus module page.)

CMS sites

The current SMS web site at http://www.maths.qmul.ac.uk is a Drupal site. I primarily maintained the undergraduate pages and the undergraduate links on the home page until very recently (May 2016) I handed over primary responsibility to a colleague. I originally wrote most of the undergraduate content and FAQs is a particularly good example of a page that I originated and have maintained ever since. Over the past few years I have moved all the information for current undergraduates into QMplus and all that is left on the SMS web site is information for prospective undergraduates.

I primarily maintain three publicly accessible "course areas" in QMplus (Moodle) – the School of Mathematical Sciences landing page, Public undergraduate information and MTH4105 - Introduction to Mathematical Computing - 2015/16 – and two "course areas" that are not publicly accessible – the School of Mathematical Sciences staff intranet and Internal undergraduate information. I maintain my Queen Mary web home page and a few other web pages linked to my home page in QMplus Hub (Mahara) and I have developed this CMALT e-portfolio in QMplus Hub.

When I stopped running my own web server on centaur I needed a home for information which was not specific to Queen Mary and which I wanted to continue to publish in the long term without relying on Queen Mary facilities. To do this I set up a new web site hosted by Google sites at sites.google.com/site/fjwcentaur, which I called "New Centaur". (At some future date I will probably drop the word "New".)

Reflection

My choices of technology as described above were determined to some extent by what was available at the time and, for example, changes of policy at Queen Mary forced me to work first with conventional web sites, then with a Dreamweaver site and then with CMS sites (Drupal and Moodle). On the other hand, I had complete freedom to choose how I ran my web site on centaur, whether to use Mahara and whether to use Google sites. Regardless of why I used them, I believe that I have used enough different technologies to have developed a good overview of their constraints and benefits.

Running your own web server is great because you have complete freedom to choose all the technology you use but you need a computer with a fixed IP address and a DNS entry, which normal domestic broadband contracts do not provide, so it is not straightforward or cost-free to run a private web server. You also need any firewalls to allow access to the necessary ports and protocols, so you need the agreement of your network managers, and at Queen Mary they are increasingly concerned about security. This was essentially why it ceased to be feasible for me to run a web server on my office computer. There is also less need now to run my own server since there are free hosting services available, unless I want to run non-standard software like the REDUCE demonstration that I used to run. I don't know how I would run something like that now but I would consider a cloud host such as Amazon Web Services.

A conventional web site with ftp access (such as provided by SourceForge) has the advantage of almost no constraints and the disadvantage of almost no structure. This means that you have to know what you are doing, which makes it very hard for a beginner to set up a new web site. This environment makes sense for the web sites hosted by SourceForge, which are intended to be used only by software developers who can be assumed to be reasonably competent at web technology. The only important conventional web site that I still maintain is the REDUCE site on SourceForge.

Sites that are essentially conventional but are maintained using a specific application, such as Adobe Dreamweaver or Microsoft FrontPage, have superficial advantages. They offer structure to the web site via templates that are maintained within the application and automatically applied to web pages, they offer support for navigation links and they hide much of the underlying file structure. They also provide good WYSIWYG editing and make it unnecessary to be familiar with the underlying languages: HTML, CSS, JavaScript, etc. However, the application is used only by agreement and there is no way to prevent direct access to the files underlying the web site. As soon as this happens, either by accident or intention, the web site structure starts to disintegrate and can be very difficult to restore. This is exactly what happened with both the Dreamweaver and the FrontPage sites that I maintained. In the case of Dreamweaver, I learnt from colleagues that not everyone who edited the web site used Dreamweaver, so they edited the files directly and broke the template structure. In the case of FrontPage nothing should have gone wrong because I was the only user, but somehow I inadvertently broke the template structure. I decided to move away from using FrontPage because Microsoft dropped it as a product and it no longer supported current versions of HTML and CSS.

A disadvantage of CMS sites is that they can only be edited online, which requires a reliable and sufficiently fast internet connection. If the internet goes down before you have saved an edit then you are likely to lose it. Hence, I tend to save my editing every so often and then re-open the editor, although this process is more disruptive than just saving a file in an offline editor. I also run a Firefox add-on to recover input (previously "Lazarus: Form Recovery" and currently "Textarea Cache"), which has saved me from losing work on many occasions (and is a good reason to use Firefox, or at least not to use Microsoft browsers). Online editors tend to be less responsive than local editors and are not WYSIWYG. For this reason, I occasionally copy and paste between an online editor and a local editor if I am trying to do something non-trivial, such as editing complicated JavaScript, for which online editors seem to provide no useful support at all.

Probably the main advantage of CMS sites is that they (should) provide all users with broadly the same experience, regardless of platform. This was one of the main reasons for moving to a CMS for the SMS web site, because it needs to be edited by a lot of different staff using different platforms and having very different levels of expertise. CMS sites also enforce a consistent look and feel to a web site. When staff in SMS maintained their module web pages using a conventional web site there was huge variability among the quality of the pages. A few looked very good but most didn't. I learnt that my colleagues have a wide range of competence and interest in web editing. Now that all our module web pages are in QMplus they are more uniform. There is a case for requiring and trying to enforce greater uniformity but there is also a case for allowing innovation and creativity, and it is difficult to know where to draw the line.

A CMS site could provide version control and backup, and page preview. Our Drupal site does this, whereas the other CMS sites I use do not (as far as I am aware). An advantage of a conventional web site is that you can provide whatever version control and backup you want, as for any other files, and some editors provide support for this. For example, Emacs provides very flexible support for version control and backup and Dreamweaver provides support for Apache Subversion. Offline editors can provide genuine WYSIWYG editing by using exactly the same style files as the web page will use when displayed on the web, and it is possible to run a minimal web server that only serves to its local host computer so that one can preview an edited page via several web browsers before publishing to the real web server.

A major constraint imposed by a CMS is that it is not normally possible to access the HTML <head> element. In fact, it is not possible to access the whole <body> element either and one can typically access only specified containers (<div> elements) within the body. This makes it difficult to change the default styling other than by using inline styling (i.e. the style attribute of HTML tags). For styling a table with many cells or a list with many items this can involve a lot of code, which can be difficult to maintain. One reasonably convenient solution is to use a few lines of JavaScript, provided the CMS allows it; Moodle allows <script> elements (and I use them quite a lot) but Mahara does not. I return to my use of JavaScript below. HTML5 defines a <style> element (with the scoped attribute) that can be used within the <body> element, but even when a CMS claims to serve HTML5 (i.e. web pages that start with <!DOCTYPE html>) its editor may still remove <script> elements, as does the SMS Drupal installation.

Microsoft PowerPoint

Description and Reflection

I normally use Microsoft PowerPoint to prepare and display slides for presentations. An alternative, as I learnt from my colleagues, is to generate a PDF file from LaTeX source with a suitable package, such as Beamer, and display it using Adobe Reader. Appearance is very important in presentation slides and in my experience the easiest way to control this is to use a WYSIWYG facility for constructing the slides. Therefore, I prefer to use PowerPoint. When I have used other, non-WYSIWYG facilities, I have found myself spending a lot of time after I have written the content changing page breaks, font sizes, etc. to make the presentation look right. With PowerPoint, content and appearance go together in a more natural way.

One reason why mathematicians prefer to use LaTeX/PDF for presentation slides is that it is easy to include well formatted mathematics. Indeed, until perhaps 10 years ago, it was difficult or impossible to include well formatted mathematics in a PowerPoint presentation. Microsoft Equation Editor provided support for elementary mathematics, but the results were not very good and for non-trivial mathematics it was necessary to produce image files of the mathematics using some other application and insert them into the PowerPoint file. If a presentation involved a lot of advanced mathematics then it was easier to use LaTeX/PDF. However, with recent versions of PowerPoint this is less true and support for mathematics is very much better.

It is easy to animate PowerPoint slides so that new information, such as the next item for discussion or the next slide, appears in engaging ways, although I think it is important not to over-do this. I believe that animation within a slide is generated by the PowerPoint application when the slide is displayed, triggered by information in a PowerPoint file. By contrast, Beamer produces animation by generating multiple slides that appear to produce roughly the same effect but at the price of large PDF files, as I learnt when colleagues have shared such files with me.

A nice feature that PowerPoint offers is the facility to write notes about each slide, which can be printed separately or together with the slides. The notes can elaborate on the slides to provide a reminder for the presenter and further detail for a reader. PowerPoint presenter mode is available when a computer running PowerPoint is connected to a secondary display, such as a data projector. It shows only the current slide on the secondary display but on the screen for the presenter it shows each slide together with the notes, a timer and various controls.

I have also used PowerPoint to prepare displays that I do not intend to present but which I want to have the structure of a sequence of independent slides rather than a single document. However, you must take care when putting PowerPoint presentations on the web because you can't assume that all viewers can open PowerPoint presentations satisfactorily (because they don't have an appropriate version of PowerPoint or a compatible application). I have recently started to adopt the following approach. In order to make PowerPoint presentations viewable as web pages, I upload them to OneDrive and then use public access links, which open the files in PowerPoint Online. This seems to work well and does not require the viewer to have PowerPoint (or any other Microsoft software).

Evidence

I used PowerPoint for my presentation in the Goldsmiths' Company Mathematics Course that I ran in 2009 and, as I described above, I published my presentation as a PDF file containing both the slides and notes, which can be accessed via the Computing in mathematics link. My original PowerPoint file can be accessed via this directory listing. I did much the same in 2010.

For several years up to 2013, I gave a Saturday morning "masterclass" for year-10 pupils from local schools on magic squares based around a PowerPoint presentation, which for the purposes of this portfolio I have published on the web using PowerPoint Online.

I gave a 10-minute PowerPoint presentation on Managing group assessment and feedback via QMplus at the Queen Mary Learning and Teaching Conference in January 2016 in Workshop 4: Using online tools for assessment and feedback.

I revised the set of "you said, we did…" PowerPoint slides prepared by our Student Support Officer, based on a generic Queen Mary version, and linked both my version and the generic version to the School of Mathematical Sciences landing page. These presentations are intended to provide students with feedback from last year's National Student Survey. The links to the two PowerPoint presentations look like this: "You said, we did… Maths / QMUL". (These two links work although the published links are no longer available). Comparing the two presentations, you can see that I modified the PowerPoint template to include "School of Mathematical Sciences" and to make "National Student Survey" and the two images at the bottom into hyperlinks, which seemed appropriate for use on the web.

Adobe Portable Document Format (PDF)

Description and Reflection

PDF is my second-choice format for publishing documents on the web when I don't feel that HTML is appropriate. I normally write documents that I plan to publish as HTML directly in HTML but I use PDF as a secondary format to publish documents that I write in some other source format. Documents I publish in PDF are primarily mathematical documents that I normally write in Maple these days, but include document that I write in other applications such as Microsoft Word. I prefer to use PDF because it is very portable and PDF viewers are available for free for every platform I am aware of, and because PDF is primarily a format for publication rather than editing (although it can be edited to some extent). Not everyone can reliably open Microsoft Office documents, but "PDF is now an open standard maintained by the International Organization for Standardization (ISO)"; see About Adobe PDF.

PDF is easy to generate from any application that supports printing, at least on Microsoft Windows. Recent versions of Microsoft Office applications intended for document preparation (in particular, Word and Excel) include a facility to export documents as PDF, as does Maple, and Windows 10 includes a PDF printer driver called "Microsoft Print to PDF". However, I generally prefer to use the Open Source PDFCreator application, which uses Ghostscript internally. PDFCreator generally produces smaller PDF files that display better than those generated in other ways. Something that I learnt from colleagues who use Linux is that PDF files generated on Windows can look very strange on other platforms because of font incompatibilities. I find that PDFCreator generates more portable PDF than do Microsoft PDF generators.

I find PDF to be a very convenient format for annotating documents, with no chance that annotation and editing get confused. Text annotation works well on a conventional computer and ink annotation also works well on a tablet with a stylus interface, such as the Samsung Galaxy Note. This means that students can, if they want, annotate PDF documents in much the same way that they can annotate printed documents for study and revision purposes.

Evidence

My web sites are all littered with PDF files and I often provide documents in both their original source format (PowerPoint, Word, Maple, etc.) and as PDF. For example, on the web site for my Goldsmiths' Company Mathematics Course I provide PDF files created by me and the other presenters and on the QMplus course area for my module MTH4105 - Introduction to Mathematical Computing - 2015/16 I provide most of my teaching materials in their original Maple format and as PDF. My personal Google web site provides my Brazil Lectures on Computer Algebra as PDF files, which I generated from LaTeX source files. (Maple did not support writing mathematical documents when I wrote these lecture notes in the early 1990s.)

XML

Description and Reflection

XML stands for eXtensible Markup Language and is related to HMTL in that both evolved as simplified versions of SGML (Standard Generalized Markup Language). In fact, XHTML is a version of HTML that is also a special case of XML. XHTML has a slightly more rigid syntax than HTML, which I prefer, so I always try to write XHTML. But XML is much more general than HTML and can be used to describe almost any information, ranging from information that is essentially prose to information that is essentially tabular. A well-designed XML data file describes itself and can be used in many different ways, including ways that were not anticipated when the format was designed. For some years, our undergraduate module details were maintained in a system that consisted of a single data file that was encoded using a bespoke language based closely on LaTeX (because the primary output format was on paper). The data file was processed by a perl script that could output LaTeX for printing, HTML for displaying on the web, and other formats that later ceased to be useful. When the original developer of this system left Queen Mary, it became increasingly difficult to maintain so that it met current requirements. In particular, the web output that it generated began to look increasingly old-fashioned and limited in applicability. At some point, I took this system over and updated it quite a lot. But I realised that it needed to be replaced by a more robust system that would be easier to maintain and develop, so I changed to a standard data format and processing software. Printed output ceased to be required, so the main requirement was to generate good HTML. Hence, using XML seemed the best way forwards.

I designed an appropriate XML format and an associated DTD. I also developed a number of XSLT files to transform the XML data file into various HTML files. Initially, this consisted of a file containing a table of modules and their key information plus one file per module containing all the information for that module. I later added another file consisting primarily of a table that summarised the overlaps, prerequisites, and other restrictions for all modules. It was quite easy to extract this data from the XML file even though it was not part of the original design. Partly for speed of browsing and partly to avoid browser incompatibility problems, I processed the XML offline every time I updated the data file. I used a standard Open Source XSLT processing application (xsltproc) that I could run on our Unix-based departmental computers and also on my Windows-based office computer under Cygwin. I was provided with access to the School installation of Apache Subversion so that I could have file backup and version control for this key School data. However, I was only provided with local access to Subversion on the School computers and not on other computers, which was awkward given that I preferred to work on my office computer and my home computer. I synchronized files between my office and home computers using Windows Live Mesh, which evolved into Microsoft SkyDrive and was then renamed to Microsoft OneDrive. I was able to access the filestore on our Unix-based departmental computers from my Windows-based office computer using Samba, and with some care I could keep all my files synchronized, but it was complicated.

I eventually moved to using a Moodle database for our module details, which I will describe later in this portfolio. One reason was the complexity and general inhospitality of the computing environment that I needed to use to maintain the information on our School web server. Another was changes within IT Services at Queen Mary, which meant that the School was scheduled to lose its local IT support. A key reason for moving to a standard XML-based system was that I hoped that our IT staff would be able to help maintain it, but that began to look unlikely. I anticipated (and still anticipate) that our module details would need to be maintained by members of our Professional Services staff in the longer term and I had already learnt that a Professional Services colleague was unable to use another (much simpler) system that I had set up on our Unix-based departmental computers because it used a command-line interface. So, even though my XML-based system was far better than its predecessor – it was more flexible, easier to maintain and produced better output – I realised that it did not have a long-term future and a much more point-and-click oriented system would be required, even if it was technically inferior.

Evidence

The output of my XML-based system can be seen for academic year 2013–14 and examples of all the data files can be accessed via this directory listing. A lot more details about the system can be found in this ReadMe file, which I wrote when I first developed the system to remind myself and any subsequent maintainers how it worked.

JavaScript

Description and Reflection

I use JavaScript quite often in web pages that I maintain, usually together with the jQuery library, which I find makes interacting with the HTML document object model (DOM) much easier, and occasionally I also use the jQuery UI (user interface) library. Although use of CSS and JavaScript in web pages is arguably not strictly necessary, it has become standard practice and can enhance a web page considerably. Appropriate formatting makes a document much more appealing. For example, a document that is clearly broken up into sections and paragraphs is much easier to read than a large document written as a single paragraph. Site-based default formatting of a web site is a good idea in order to give a sense of consistency, but writing default styling that works correctly in all situations is complicated and, in my view, default styling is not always appropriate. For example, paragraphs within list items might not be appropriately separated, and whilst not using table or cell borders might be an appropriate default, some tables are much clearer if the rows, columns or cell are separated by lines. Moreover, allowing some user interaction, such as collapsible sections, can make the structure of a document clearer and complicated structure less intimidating to the reader.

For conventional web pages I use JavaScript to provide dynamic effects, such as collapsible sections, table sorting, freezing table headings and dynamic highlighting. For CMS web pages, I use JavaScript to provide dynamic effects that are not provided by the CMS, much as described above, and also to work around limitations imposed by the CMS. In particular, a CMS does not provided access to the HTML head element and hence not to style elements. (To be precise, this was the case before HTML5, which does allow style elements within the body element if it has the scoped attribute, but even so a CMS may not allow it.) Without access to style elements, it is necessary to specify the style of every element inline by using its style attribute. This can get tedious, especially in large tables or long lists, and can be very difficult to maintain, although some online editors provide quite good support for styling table elements.

I often find it easiest to style multiple elements with a line or two of jQuery, which is quick to write and easy to maintain. Fortunately, script elements can appear in the HTML body element and it often works well to manipulate DOM elements by including a script element immediately below the DOM element. Indeed, it is often recommended to put JavaScript at the bottom of a web page, if this is feasible, so that processing it does not slow down the initial rendering of the page. When this doesn't work it is easy enough to force the JavaScript to be run only after the document has fully loaded and jQuery provides a very elegant mechanism for doing this.

It is probably more elegant to put as much styling as possible into style elements and it may be more efficient, but in practice I have not noticed any slowdown caused by running fairly large JavaScript elements. Many web pages already include huge amounts of JavaScript by default, so a relatively small addition by me is unlikely to make much difference. One issue I have noticed is that errors in the default JavaScript can prevent my JavaScript from running. This was the case in the past with QMplus and I found that it was critical where I put my code in order to ensure that it ran. (The buggy QMplus code now runs in a separate file and no longer seems to interfere with my code.)

A CMS may not explicitly support the use of JavaScript, in which case my solution is to embed the JavaScript in a convenient HTML element, which generally works well in Moodle. This allows default elements of the page to be changed, which makes Moodle very customizable.

However, not every CMS allows users to include JavaScript, and I have just discovered that QMplus Hub (Mahara) appears to delete any script element that I try to include in this portfolio. The MOS badge that I have displayed in my overview of this section is available in various ways. I initially tried to use the embed code, which consists of a <div> element followed by a script element. This didn't work because when I saved the edit QMplus Hub deleted the script element. (I have had the same experience when using the WYSIWYG editor in Drupal, but switching to a different editor solves that problem.) I therefore proceeded in a more conventional fashion and downloaded an image file for the badge, uploaded it to QMplus Hub, copied the URL of the image file, inserted an image element essentially by hand and then wrapped the image in an anchor element. I presume the result is more or less the same as would have been produced by the embed code if it had worked, but the embed code referenced a file of JavaScript that I have not looked at, so I can't be sure.

Evidence

You can look at the source code of any of the HTML pages I reference below to see exactly how I implemented them, so and I won't describe my implementations in detail.

I originally implemented a facility for displaying full details of our undergraduate modules as conventional web pages, which still exist for three academic years at www.maths.qmul.ac.uk/~fjw/modules-2011-12, www.maths.qmul.ac.uk/~fjw/modules-2012-13 and www.maths.qmul.ac.uk/~fjw/modules-2013-14. At the top of each of these pages is a accordion widget that I implemented using jQuery UI. Below this is a list of all our undergraduate modules that can be sorted in situ. I implemented the sort facility using a jQuery plugin called tablesorter. If you scroll the list you will notice that instead of scrolling off the top of the window, the table head detaches itself and remains fixed at the top of the window until you scroll up again, when the table head rejoins the table body at the appropriate point. I wrote the code to implement this myself because I couldn't find anything that did what I wanted. It was tricky to get the code to work correctly and interoperate correctly with tablesorter in all possible window configurations. It's still possible to corrupt the table head by appropriate (or inappropriate) combinations of changing the window size and scrolling, but I decided it was good enough to be usable. If I had continued with this technology I might have fixed it properly by now!

Clicking on a module code takes you to a single page of details for that module, which consists of another accordion widget that I implemented using jQuery UI. Finally, expanding the central tab at the top of the main module list provides access to several links, one of which provides a table that summarises the overlaps, prerequisites, and other restrictions for all modules. In this table, the row under the mouse pointer is highlighted to make it easer to read across all the columns of a wide table, which I implemented using just CSS, and clicking on a row toggles a darker highlight, which I implemented using jQuery, so that it is possible to have a primary and secondary row highlighted.

On the QMplus page for the module MTH4105 - Introduction to Mathematical Computing - 2015/16 that I have taught for the past four years, I use jQuery to change the icon associated with Maple files from the default unrecognised file icon to a Maple leaf icon. A few years ago, at my request, our E-learning Unit designed a nice Maple leaf icon for use with Maple files and incorporated it into a standard facility for Adding Icons to Your Pages. However, what I really wanted was that the standard facilities for adding file resources to QMplus pages would recognise the file type as Maple and use the appropriate icon. (I think that the choice of file icon should be table-driven and easily configurable on a per-site basis, which appears not to be the case at present.) My solution was to add a few lines of jQuery to my QMplus page that change the icon when appropriate. I would have liked to drive this by the filename extension, but this does not appear to be exposed on the web page. Therefore I developed a convention that the resource name should include the text "(Maple)". This allows me to select the right icons to change and the mechanism seems to work well.

Very recently, I set up a QMplus assignment manager for staff to upload files. Even though I set the grade type to none, the assignment manager still refers in several places to grading, which is not appropriate for my purposes. By examining the HTML code for the page, I established that the references to grading all occurred in cells in one or two tables (depending on the user's role). I concluded that the assignment manager would be more acceptable to staff if I deleted any table rows with cells containing the word "grading" or "Grading". I achieve this by adding the following code in a script element to the assignment manager description:

// Use jQuery to remove table rows that refer to "grading" or "Grading":
$(function() {
   $("table.generaltable td:contains(rading)").closest("tr").remove()
})

This example is fairly typical of the way I use JavaScript to tweak QMplus.

Description

Background

QMplus is the name given to the Moodle installation used by Queen Mary. I have been the QMplus Administrator for the School of Mathematical Sciences (SMS) since this role was introduced in summer 2015 although I had been responsible as the SMS E-learning Coordinator for most aspects of the use of QMplus by the School ever since Queen Mary decided to switch from Blackboard to Moodle in around 2010. A few years before that, I and a couple of colleagues had begun to use Blackboard instead of our School web site for some student information. I did this partly to gain some experience with a VLE, since the Science and Engineering Faculty generally had very little involvement with VLEs at that time, and partly as a way to solve the problem of how to make sensitive information available to students via the web in a secure way. I set up a Blackboard page that required login and was accessible only to Mathematical Sciences students, and used it to provide information about our Student-Staff Liaison Committee and some personal information about tutorial arrangements and results that should be accessible only to the data subject. The first involvement of the School with Moodle was migrating this handful of Blackboard pages to QMplus for academic year 2011–12. I migrated my page myself by deciding what kind of Moodle resource to use to represent my resources in Blackboard and then copying and pasting.

By academic year 2012–13 the School had 18 course areas in QMplus. During that year, I liaised with the E-learning Unit to migrate the content from all our remaining conventional module web pages into QMplus course areas, which was quite tricky because there was no uniformity at all among these pages. Science and Engineering Faculty policy was (and still is) not to be too prescriptive about the use of QMplus, preferring to make the transition as seamless as possible for teaching staff, given that there was some lack of enthusiasm for the transition. (Humanities and Social Sciences Faculty policy is much stricter, possibly in part because HSS has used VLEs for much longer.) Teaching staff, especially in Mathematical Sciences, felt that they were being forced to use new technology that provided no advantage over the conventional web site that we had used successfully for the past couple of decades. (We were early adopters of the World-Wide Web and had run our own servers within the School for a long time.) I proposed a timetable for migrating teaching materials and I proposed roughly how material should be migrated. I did this by collaborating with the E-learning Unit to design a template into which our teaching materials would be copied and annotating the template to indicate where to find the information to populate each component. The migration was performed by "e-learning assistants", who were Queen Mary students trained and paid by the E-learning Unit.

This approach to migrating our teaching materials got the job done. My view was that this step was primarily to deliver teaching materials to QMplus and that module organisers should then use those materials to build their course areas. However, in most cases that did not really happen, and many course areas still look like a pile of building materials. I don't know whether this is because some module organisers think this is appropriate, or because they don't realise that alternative arrangements of materials might work better, or because they don't want to spend any time on this aspect of their teaching. It would have been better if each module has been migrated with more care, but that would have required a lot of time and expert input and the resources were simply not available. Schedules slipped and the migration had to be completed quickly to be ready for the next academic year, so there was no time to consider what went where in any detail. Since the migration, I have tidied up any major problems that I have noticed, but I am reluctant to interfere with anything that may reflect a module organiser's preference just because it's not my preference. The best time for such tidying up is over the summer between academic years, especially when a module organiser changes.

By academic year 2013–14 the School had 70 course areas in QMplus, which included all our taught modules. I created a second generic QMplus course area to accommodate more of the student information from our web site. This site is open-access whereas the first generic site I created is not. Over the past three years, I have moved all the information for current undergraduate students from our web site into QMplus. I finished this move quite recently (March 2016). Another development for academic year 2013–14 was the introduction of landing pages for faculties and schools. I was not directly involved in the creation of the SMS landing page but I have been the main (and I think only) editor since it was set up. I added an image (a graph in three dimensions of the complex gamma function) and I have subsequently changed the top half of our landing page quite a lot.

Our students complained that it was rather boring, so I replaced the brief description of the School by three pithy quotes from famous mathematicians with pop-up titles to indicate who was quoted and provide links to more details. I added links to interesting mathematical background information such as the "Theorem of the Day" web site. After a policy change by the E-learning Unit on the structure of landing pages, I changed the image at the top of the page to the current version. This is based on an image that appeared on our School web site for a while in the past and which I liked. I think it was created by the Queen Mary web team. I retrieved it, I think, using the Internet Archive Wayback Machine since I could no longer find a copy locally. I then trimmed it to the right shape and so removed most of the part that related to our Astronomy Unit, which has moved to what is now the School of Physics and Astronomy. I then rescaled the image to have exactly the required pixel dimensions, which meant scaling up the number of pixels by a small amount. I thought that this might spoil it, but in fact I think it looks quite good. Quite recently, I wrapped anchor tags around the header image and text that link to the SMS web site.

Our landing pages are quite tricky to edit because they involve HTML <div> elements nested several levels deep that are positioned using CSS to give the grid structure. The graphical editor provided in QMplus does not support this layout, which is invisible and so very easy to mangle. Hence, it is more or less essential to edit at the HTML source code level, but the only reliable editor currently provided in QMplus provides no syntactic support whatsoever. This makes it tricky to keep track of the <div> nesting, and it is still quite easy to mangle the code, especially when trying to move large chunks. Every so often I discover when viewing the page normally that I have mangled it and I have to go back into the editor and fix it. When an edit is saved, the resulting code is run through a syntax checker (probably a version of Dave Raggett's HTML TIDY program or something similar), which tries to correct any invalid syntax. In particular, it will delete a closing tag that has no matching opening tag and insert missing closing tags, but sometimes it inserts them in the wrong place. There can be interesting consequences when a closing <div> tag is inserted in the wrong place in a nested <div> structure, and it can be inserted a long way away from where the code was edited such as near the bottom of the file below the last closing <div> tag!

Another interesting peculiarity is that occasionally editing our landing page breaks our embedded Twitter feed. The reason is that the editor or the syntax checker converts

<script type="text/javascript">

to

<script type="mce-text/javascript">

and the browser then ignores the script element. Now that I have worked out what causes the Twitter feed to stop working it is easy to go back into the editor and fix it, but I need to remember to check the Twitter feed after I have finished editing the landing page.

Staff intranet

A couple of years ago, I set up a QMplus course area to act as our staff intranet. Initially, I was not enthusiastic to get involved in this project although I thought it was a good idea. Two of our Professional Services staff requested that a course area be created for this purpose and (presumably) intended to set it up and transfer content to it, but didn't. Somewhat later, one of them asked me to add some information (a new set of email lists for groups of School staff) to our intranet. She may have intended that I added this information to our old intranet because by then I was one of very few staff who maintained it. But I decided it was time to start using the new intranet site in QMplus. It had been created using our undergraduate module template, which was quite inappropriate, so I reconfigured it, primarily by removing the information added by the template. I then added the new email lists as the first content to the new intranet and advertised it to School staff. After that, I put any new information intended for our intranet onto the new QMplus intranet.

Some time later, when I had more time, I began moving content from the old intranet to the new intranet. I had access to the underlying filestore, but it contained a lot of redundant files that were no longer linked to the index page but had not been deleted. So instead I downloaded the index page and all files linked to it by using the DownThemAll! Firefox add-on. I set up a number of sections on the intranet page to reflect an updated version of the structure of our old intranet and then uploaded the files from the old intranet into appropriate places. These files were roughly half HTML and half PDF. For each of the HTML files, I created a new (Moodle) page and copied the main content from the old intranet file into it, excluding the old Dreamweaver template content. In some cases, especially when I had written the original file, I updated the content to correct trivial errors like broken links, whilst trying to avoid inadvertently rewriting School policy!

The PDF files will be more difficult to update, especially those that I didn't write and hence don't have the source for. The old internet is now essentially frozen and will at some future date disappear. The new QMplus intranet has a number of significant advantages: it is accessible from anywhere, not just from computers on the School network; it can be edited by anyone who is given the appropriate access, which any member of staff who already has appropriate access can grant without needing to ask ITS to do it; it is now easy to write new documents as QMplus pages, so I hope that we will make less unnecessary use of PDF documents, which are harder to maintain.

It seems inappropriate to have the editing role on our staff intranet called "Teacher" so I have renamed that role to "Editor". Currently, most staff access the intranet using the Viewer role that all staff have for the whole School of Mathematical Sciences category. However, I recently set up an assignment manager on the intranet to allow staff to upload PDF copies of their exam papers ready to be made available to students next academic year. To make this work, I needed to give examiners the "Student" role, which I renamed to "Examiner". It was fairly easy to run down the list of enrolled users and add the Examiner role for each member of staff who is currently an examiner.

QMplus maintenance

I control the "rollover" of our QMplus course areas for the next academic year, which means erasing current student data and reconnecting course areas to the new student module registrations. Before last summer, I had to do this separately for each course area whereas now there is a category level facility to initiate rollover, which is less work. One of the motivations for introducing the QMplus Administrator role was to provide a role within academic Schools that would have access to this category-level rollover task. It is important to roll a course area over at an appropriate time, taking account of the teaching cycle for each module.

My strategy is to roll all our undergraduate modules over around 1^st September, which means that nothing changes before our late summer resit exams in August. However, some of our postgraduate modules need to be rolled over much later, since MSc courses run for a full twelve months and students submit their dissertations at the end of the summer. MSc exam boards meet after the end of each academic year and it can be useful to keep various MSc student submissions available in QMplus until after the exam board meetings, hence some of our MSc course areas are not rolled over until October or November.

I currently find it convenient to organise our QMplus module course areas as three categories:

• modules that have an undergraduate version (which is most of our modules);
• modules that are only available to MSc students;
• support information that is not specific to any module.

Hence, I maintain three subcategories under the main School of Mathematical Sciences category, which I call "SMS UG/PG Modules", "SMS PG Modules", and "SMS Support". All our QMplus course areas are in one of these three subcategories except for our landing page, which is in our main category. I have revised these subcategory names a few times. They need to be reasonably descriptive but not too long since they appear in the breadcrumb navigation. As our MSc programmes have developed over recent years I have modified our category structure to accommodate them in the best way possible. The current categorization is based primarily on rollover requirements:

• I roll over the SMS UG/PG Modules pages around 1 September;
• I roll over the SMS PG Modules pages when the appropriate PG staff ask me to;
• SMS Support pages need ad hoc treatment and may not need to be rolled over.at all.

As our QMplus Administrator I also have (limited) access to the QMplus Enrolment Mapper. By default, if a QMplus course area for a module is set up correctly then it automatically enrols students who are registered for the module in our Student Information System (SIS). The QMplus Enrolment Mapper allows alternative and much more flexible enrolment rules to be set up and managed. My access allows me to do everything I could reasonably want to do for course areas in the SMS category except delete mappings. (This is frustrating. I can create new mappings, edit them and disable them, but if I want to delete a mapping I have to ask the helpdesk to do it for me.)

Most of our level 6 and 7 modules have two or more versions to allow them to be taken under slightly different assessment regulations. This means that they have different module codes and are treated as independent modules in SIS, but all the students are taught together in essentially the same way, so it makes sense to maintain only one QMplus course area for all versions of the same module. This is facilitated by setting up appropriate enrolment mappings.

I have extended this idea slightly and grouped together our Third-Year Project module and our MSci Project module. These are genuinely distinct modules but they are run in identical ways and in practice they have the same module organiser(s) and the same list of potential projects. Originally, they had separate QMplus course areas, but a couple of years ago I merged them and arranged for enrolment to be mediated by an enrolment mapping rule. Enrolment for our generic course areas is also handled by the enrolment mapper using rules to enrol, for example, all first-year and second-year undergraduate Mathematical Sciences students. Our collection of enrolment rules is quite complicated and it has taken some time to get all the rules set up correctly. In future, maintenance should consist only of setting up rules for new course areas and changing rules following policy changes.

In-term assessment grades

Only a QMplus Administrator has access to the task to transfer marks from the QMplus gradebook to SIS. Until this academic year, Mathematical Sciences did not use the QMplus gradebook much. However, the E-learning Unit began development of a facility called Gradesplus that displays a student's grades in QMplus in the form of a histogram of grades for the whole module with the bin containing the student's own grade highlighted. This gives a very clear picture of a student's performance relative to their cohort and I think it is a useful addition to QMplus. I therefore asked all SMS staff to put their in-term assessment results into the gradebook and I offered that we would be one of the Schools to pilot the use of Gradesplus. In order to encourage my colleagues to use the gradebook I explained that we could only offer Gradesplus to our students if there were grades in QMplus to be displayed and I offered to transfer grades from QMplus to SIS, so that teaching staff would not be required to enter them twice.

This project seems to be working well. By the middle of the second semester, most of our first-semester in-term marks were in the QMplus gradebook and in SIS. One or two colleagues put their marks directly into SIS, in which case I copied them into QMplus. I was able to use our first-semester grades to test development versions of Gradesplus and pick up problems such as inappropriate gradebook items appearing or appropriate gradebook items not appearing. I also made a few cosmetic suggestions. When Gradesplus was first released to students, I invited our students to try it and let me have some feedback. The response was positive and the main request was that Gradesplus should work for all grades obtained over all years (whereas currently it only works for grades for the current year that are stored in QMplus).

I discovered a couple of problems with the facility for transferring grades from QMplus to SIS. It will only pick up specific gradebook items and not the course total. This is a serious problem because one of the advantages of using QMplus is when grades for several in-term assessment components need to be aggregated before transferring to SIS. The natural way to do this is to use the course total to compute the aggregate. I had to work around this problem by downloading both a CSV file in the right format for uploading to SIS, but containing the wrong grades, and a second CSV file in the wrong form but containing the course total. I then transferred the course total from the second file to the first (using the Microsoft Excel VLOOKUP function). Fortunately, I only had to do this for a couple of our modules. I have submitted a development request to fix this problem.

The other problem I discovered is that the file generated by QMplus for "Z occurrences", which correspond to associate students attending for only the first semester, contained incorrect data in one field, which prevented the upload to SIS. I isolated the cause of this problem by downloading a template file from SIS and comparing the two files. When I corrected the data by hand I was able to upload the file. I reported this problem to the ITS helpdesk, which referred it to the SIS helpdesk, who told me there was no error in the SIS data. I understand that they referred the problem back to the ITS helpdesk and I heard nothing more! I hope it will be fixed for next academic year.

Last semester, I initially wrote some very brief guidance on how to put marks into QMplus and referred colleagues to the QMplus help. This was not as successful as I had hoped and I had to talk one or two colleagues through various steps or intervene and do things for them. After that experience, I wrote more specific and detailed step-by-step guidance and relied less on colleagues reading the QMplus help. The QMplus help recommends that the best way to enter marks into QMplus is to use an assignment manager, so I duly followed this advice initially. Indeed, my own experience has been mainly with assignment managers since I use student file upload, so the official advice seemed reasonable. However, when I investigated more closely I discovered that it is actually much simpler to create a gradebook item directly and not to use an assignment manager. So, for the second semester, I changed my approach. I created a new gradebook item for all modules that needed one and I wrote new guidance on exactly how to get marks into a gradebook item, either directly or by file upload. This has been far more successful and I have had almost no requests for assistance. When I roll our QMplus module pages over I will replace any unnecessary assignment managers with simple gradebook items. I also decided that it would be beneficial to turn on the display of rank within the gradebook; it is off by default. By creating gradebook items myself I can ensure that we use consistent settings across the School.

Evidence

The current version of the School of Mathematical Sciences QMplus Landing Page and the links on it illustrate many of the issues I have discussed above. Annual snapshots over the past three years exist in the QMplus Archive but unfortunately most of this is no longer accessible from outside the Queen Mary network. Currently (March 2016, but this may not continue to be the case) the 2014/2015 Archive is accessible and the 2014/2015 School of Mathematical Sciences QMplus Landing Page is still open-access. My developments to the landing page over the past year or so are fairly clear: the top halves of the two versions are significantly different. The links also show the way I have developed my approach to linking to content on different course areas, which I describe further below.

A (PDF) screenshot of the QMplus Enrolment Mapper shows what the user interface looks like. The mappings for course code MTH717U-A* show how the semester-A and B versions of the Third-Year Project module are mapped to the QMplus page for the MSci Project (MTH717U) and the mappings for course code NS_SMS_EMP show how all first and second-year undergraduate Mathematical Sciences students are enrolled for this QMplus page, which offers optional employability skills and is not related to any specific taught module. I recently edited the latter mappings. The enrolment for this page was originally sets up (not by me and presumably by the helpdesk) to enrol the first-year students on each of our individual undergraduate degree programmes, which is an unnecessarily complicated way to do it. It is also unreliable because it would need to be updated whenever we introduce a new degree programme (and we have recently introduced two, one starting this year and one next year). The colleague responsible for the employability programme asked me to enrol all second-year undergraduates, so I took the opportunity to simplify the enrolment mapping for this page to just the two rules now shown.

A screenshot of the grades overview for a typical first-year Mathematical Sciences student illustrates how I have configured the gradebook for our modules to display a student's grade and rank for each of their Mathematical Sciences module, where appropriate. INT016 is not a Mathematical Sciences module and MTH3100 is a special case. However, this is work in progress! I have just noticed that the ranks are wrong in that the total number of students is shown as the same for every module. QMplus is incorrectly picking up the total number of students for one module (mine, because I accessed this student via my module page) and using it for all modules. There is another way of accessing what should be the same grade overview which displays the total number of students as 0 for every module!

Reflection on QMplus administration

I think that a School the size of Mathematical Sciences, with around 60 permanent staff and around 70 QMplus course areas, needs two QMplus Administrators. This would allow them to deputise for each other when one is away and provide some continuity if one were to leave the School. It would also allow them to focus on different areas, as reflected by our current category structure. Since I am also our Director of Undergraduate Studies, I naturally focus on information for undergraduate students and I feel that our information in QMplus intended only for postgraduate students needs better support. For example, the template that I developed is intended for undergraduate modules and does not entirely work for postgraduate-only modules.

A School needs at least one member of staff whose role includes QMplus maintenance and who has some technical expertise in editing web pages. This is particularly evident for QMplus landing pages, which cannot be edited naively using only the graphical editor. But I believe that making optimal use of QMplus more generally requires some knowledge of HTML and CSS at least. For example, how do you link to content within a QMplus page in a reliable manner? It is tempting to link to a section using a URL of the form

http://qmplus.qmul.ac.uk/course/view.php?id=m#section-n

as used in the menu block, but Moodle makes it very easy to move sections around and when you do that the section numbering and hence the link URL changes. So this approach is not reliable. It would be very useful if Moodle automatically generated permanent named anchors for each section, but as far as I am aware it does not. I have thought about providing such a facility using JavaScript but currently I use a simpler but perhaps less elegant solution. When I want to link to a section, I add a subtitle that includes an id attribute, as in this example:

<h3 id="modules">Modules: information relating to modules taught by the School of Mathematical Sciences</h3>

and then include the id value as a fragment identifier in the link URL. This seems to work quite well, except that the subtitle rather than the section title appears at the top of the target page and the use of subtitles is somewhat artificial. So I regard this as something of a hack. In order to make this work at all I had to change the course format from collapsed (the default) to uncollapsed topics because it doesn't make sense to link to hidden content. The facility I would like in Moodle is similar to a facility that I discovered in Mahara (by reading my page source code), namely that Mahara gives each distinct block a unique and permanent id attribute of the form

blockinstance_123456

I use these as fragment identifiers to link to evidence within the CMALT Appendix page of this portfolio, for which I use a single-column format so that documents and images are large enough to be readable on smaller displays.

Another reason to change our two generic UG information pages to uncollapsed format, which I did fairly recently, was that I was getting complaints from both staff and students that they could not find information. I concluded that one reason for this was that I often linked from our landing page directly to specific resources, such as our module database, rather than to the page or section containing the resource. I had assumed that users would use the breadcrumb menu to explore the page containing a resource, but apparently not. Now that I link to sections rather than specific resources, users are forced to look at all the resources and pick the right one, thereby also becoming aware of the full range of resources available.