Scandinavian-Canadian Studies

In several documents, there are dates encoded in two slightly different ways:

<date value="1998">1998a</date>
<date value="1998a">1998a</date>

These are actually both from the same document. First, check whether this variance actually affects the output in any way; following that, pick one format and regularlize all instances.

Changed the default sort in the document index pages to "volume" rather than title, title being rather meaningless. I also added a secondary sort key of "start-page", which means that the documents sort themselves in the order they actually appear in the volume (once the page numbers have been fixed by the print volume release).

Then I implemented a TitleSortComparator, with the eventual intention that it be used to sort the titles, excluding leading articles. It works, but I'm unable to use it on this particular project right now because all the XSLT is version 1.0, which doesn't allow the calling of an external collator. However, it's been implemented and tested on the Mariage project, and more extensively documented there.

Logged in to John's account on Verio to change the DNS for scancan.net to UVic's servers (dns1.uvic.ca and dns2.uvic.ca). Then Greg wrote to sysadmin to get the DNS entry added.

Then I imported all the old tables of contents for the 14 volumes preceding our electronic issues, and made a page from them. The process was horribly tedious -- the old data was in a completely idiosyncratic set of tables on the old Web site, and it took a lot of hacking to get it into well-formed XML, then write XSLT to convert it to a usable format.

Met with John for an hour and a half. We did a number of things:

• Got the URL http://scancan.uvic.ca set up and working (at least, Greg got it done at our request).
• Researched available domain names and found scancan.net is available, so John registered it through Verio. When the confirmation details arrive, we'll be able to point the various DNS servers correctly and we'll have that working too.
• Looked at a range of different indexing services from Appendix 3 of the document mentioned in the post below. Most seem to be thoroughly commercial, and none had good info on how to register your journal or get it indexed, except the MLA, which requires paper submissions. Wrote to a couple of them, including the MLA, for more info. We should get the journal indexed in two or three of these if we can, but we want to avoid giving up the right to serve it open-access ourselves.
• Discussed the SSHRC grant application again, in a little more detail.
• Discussed the possibility of making the old editions available online as PDFs. John has Word files, and we could use those to generate the PDFs as long as they're complete, correctly paginated, etc.

John has expressed a desire that the TOC pages display the article titles sorted alphabetically, but without the articles included. We now have a way to approach this, using a custom sort comparator written in Java, as we've done for the Moses sort. I wrote a comparator which strips articles before comparing titles, and tested it with JUnit tests, then uploaded it to the Cocoon WEB-INF/lib directory and restarted Cocoon. Then I came up against the same problem as I had with Moses: the XSLT is all version 1.0, so Saxon 8 isn't appropriate for it. I did a bit of testing with by creating test pipelines using 2.0 versions of the files, but nothing works properly (XIncludes are not XIncluded, and the sort doesn't appear to work either). If this sort is really important, and can't be achieved simply by adding a new version of the title to each document, then we'll have to rewrite all the XSLT, but it doesn't seem worth it at the moment. The comparator will be very handy later when we work on teiJournal, which will need it, so it's not wasted work.

John has been studying the turgidly-named by quite interesting Study of the Feasibility of Open-Access Publishing for Journals Funded by SSHRC's Aid to Scholarly and Transfer Journals Programme (PDF warning), which I also took a look at myself, with a view to ensuring we're on the right path and taking the right decisions, and looking at the possibility of SSHRC funding for ScanCan in the future. We do seem to be on the right track, although there's a lot more we could do in terms of:

• Specification of licensing and copyright terms up-front for authors (probably implying the adoption of a Creative Commons licence). We already knew about this.
• Institutional archiving of the journal. We should talk to the library about this.
• Encouraging authors to self-archive their own papers in their own archives.
• Having the journal indexed in one or more public indexing services. This would involve identifying appropriate indexing services, registering with them, and providing an appropriate API or metadata source for them to use.
• Looking at the SSHRC grant application in detail (deadline June 30) to see if we can qualify for support. The requirement for four published issues in the last two years would seem to rule us out, but we might qualify under an alternative criterion which requires twelve published articles in the last two years. If we can get funding, we can set some folks to work digitizing the old issues.

John wants two kinds of change to the TOC:

1. Ordering by title should exclude initial articles, and should use special titles for reviews. This will require that he provide me with a sorting title I can use for this purpose, so it's down to him to do that.
2. The single-page TOC to be split into Articles and Reviews.

I got 2) working today:

• There are two new items on the menu, Articles and Reviews, and each points to the same contents.htm, but with a different search parameter (type=article, type=review).
• The contents.xq code now reads that parameter, and filters the list of documents accordingly. It also includes that parameter value in a <classCode> element in the <teiHeader> of the output document, so the XSLT can find it.
• The site_doc_list_to_toc.xsl XSLT now provides a different title based on that parameter (from variables in the journal_globals.xsl file).
• It also uses that parameter as part of the JavaScript call in the table head cells' onclick events, so that the URLs called by these events now pass the type parameter back into the sitemap.
• The menu.js JavaScript has been slightly modified so that it can tell the difference between the two URLs. Previously it was discriminating on the basis of the bare filename; now it checks whether the href attribute in the link forms the first part of the current page filename, so the correct button ends up being depressed on the menu.
• The sitemap ended up not being modified at all.

This approach was about the third one I tried; it seems the most flexible (future requirements could add new values for the type attribute, and only small modifications would be needed to accommodate them), and doesn't change very much. The only oddity is the fact that on the Reviews page, which consists only of one document type ('review'), there's still a Type header/column. To suppress that would require that we check to see whether there's more than one value in the <classCode> elements in all the documents in the list. That would be easy with XSLT 2.0, using distinct-values, but it might be a bit of a kludge in 1.0, which is what this project is using. It is possible though, so we might do it in future.

Following much discussion with John, we need to split the current TOC page into an Articles page and a Reviews page. Reviews need to be sorted on some basis that isn't quite clear yet; articles need to be title-sorted without their initial articles, which might not be so easy to do in XSLT, so for both cases it seems we might require a <title type="sort">.

New review came in for vol 16 from John. Marked it up. It's heavily laden with quotes, so it took a while. Some outstanding issues (minor apparent typos etc.) on which I'm still waiting to hear back from John.

Some back-and-forth with John about bitty things in two documents, and also a tweak to the TOC XSLT code (it had no handler for <gloss> elements in titles).