Contact Us on 0800 123 4567 or info@example.com

Devonshire

An electronic edition of the Devonshire manuscript
September 10, 2007

Diagnosed another markup oddity resulting in display problems

Posted by on 10 Sep 2007 in Activity log

The display problems with Looking more closely at LDev151-TM1091 are caused by a markup oddity, which also afflicts LDev047.2-Troilus:

Both of these poems contain <note> elements as children of the <bibl> tag in the head element of the poem. However, the normal procedure is to put <note> elements as following siblings of the <bibl> element, not inside it. There are 551 instances of <note> elements appearing as following siblings to <bibl> elements inside the <head> tag, but only three instances of notes appearing inside the <bibl> tag, LDev047.2-Troilus (which contains two), and LDev151-TM1091, which has one. Since this particular structure screws up the presentation of the page, because the XSLT isn't expecting it, and since it's actually hard to rewrite the XSLT to accommodate this oddity, I've suggested to Karin that these three instances be fixed.

September 7, 2007

Fixed XSLT bug

Posted by on 07 Sep 2007 in Activity log

Found the XSLT problem; it was caused by two different abbreviation forms which are marked up in the same way, which is a problem for the markup team (and passed on to them); however, my code should have been trapping for it, and that's now working. The status now is:

  • 96 poems are functioning correctly WITH apparatus.
  • 89 poems are functioning correctly, but don't have apparatus (either because it's not done yet, or there are no other witnesses).
  • 6 poems have a problem with text appearing mingled with the line numbering.
  • 1 poem has a block of text in an odd font.
  • 1 excerpt (the last) appears to be missing, or has no title and so can't be accessed easily.

Survey of where we're at

Posted by on 07 Sep 2007 in Activity log

Following the addition of a lot of new markup in the form of apparatus files during the summer, I took a look at all the poems in the collection to figure out what's working and what's not. Results:

  • 88 poems are functioning correctly WITH apparatus.
  • 88 poems are functioning correctly, but don't have apparatus (either because it's not done yet, or there are no other witnesses).
  • 6 poems have a problem with text appearing mingled with the line numbering.
  • 9 poems don't display because of an XML/XSLT error.
  • 1 poem has a block of text in an odd font.
  • 1 excerpt (the last) appears to be missing, or has no title and so can't be accessed easily.

Small fix to Array.indexOf prototype function

Posted by on 07 Sep 2007 in Activity log

My simple function for Array.indexOf doesn't actually cover the overloaded version in the JS 1.6 standard, which has an additional start parameter. Updated it to this:

if(!Array.indexOf){
    Array.prototype.indexOf = function(obj, start){
        for(var i=(start||0); i<this.length; i++){
            if(this[i]==obj){
                return i;
            }
        }
    }
}

This is based on a tip here.

July 10, 2007

Making it work on the evil browser

Posted by on 10 Jul 2007 in Activity log

After getting the Apparatus stuff working on Firefox, I started the crossbrowser testing. Opera needed one tweak: its Array() object doesn't support the indexOf() method (introduced in JavaScript 1.6), so I had to add an implementation of that:

if(!Array.indexOf){
    Array.prototype.indexOf = function(obj){
        for(var i=0; i<this.length; i++){
            if(this[i]==obj){
                return i;
            }
        }
    }
}

Tested also on Safari 2 (Mac) and Safari 3 beta (Windows), and on Konqueror. The last has some tiny layout issues, but other than that, everything works.

Except for IE, of course. IE doesn't support element.getAttribute(), so I had to put in a few traps with document.all which call use element.className instead. But it does work!

So all that's left now is the switching between the expanded view and the old-print-style view (hereinafter referred to as "the curmudgeonly view"). That will take a little thought.

And more on Apparatus

Posted by on 10 Jul 2007 in Activity log

Further refinements:

  1. Fading and unfading of the notes dif on the right of the text, which was broken by the new apparatus code, is now fixed.
  2. Interaction with the apparatus area now automatically unfades and unhighlights all areas above, so the main text area returns to its normal settings.
  3. Highlighting of the lemma in the popped-up line has now been refined somewhat. Where the whole line is the lemma, no highlighting is done, because it's pointless. Similarly, where there are multiple instances of the lemma in the line, no hightlighting is done, because only one of the highlights would be the right one, and the code cannot know which (see discussion below).
  4. I tweaked the page layout so that the popup line appears to the right of the apparatus, and does not obscure it.

This issue of highlighting lemmas could use some input from Ray and Cara. Take the example of LDev003 line 8:

do displease the not tho I do reffreyn

One reading shows this:

8 do] omit. LDev145, AAH06, LEge20, STC13860_12.

Now, "do" appears twice in the line, so it's not clear from the apparatus which "do" is omitted here. We might be confident that readers may decide correctly for themselves (although I can't; the line seems OK with either "do" missing), but it's not ideal. I think a better solution would be to intercept these instances and rewrite them like this:

8 do displease] displease LDev145, AAH06, LEge20, STC13860_12.

(assuming that the first instance of "do" is the one omitted.) This would make it clear which "do" is dropped, and in the process it would disambiguate the reading from the point of view of the processing code, so that I could confidently highlight "do displease" when popping up the line.

More progress with apparatus rendering

Posted by on 10 Jul 2007 in Activity log

Added some refinements to the code which provides the apparatus interactivity. The key problem we're solving is the requirement that when the user clicks on the lemma in an apparatus item, the poem line should pop up, with the lemma highlighted inside it. This has a couple of implications:

  1. We can only work with the plain text of the line, not the actual HTML tags in the line, because we need to match the text of the lemma against the text of the line in order to provide the highlighting. Therefore...
  2. ...we can't show a fully rendered line (with e.g. deletions showing a line-through) because the HTML tags have to be removed.
  3. For an effective comparison (enabling highlighting), both the lemma and the line must be normalized (all spaces and returns reduced to a single space, and leading and trailing spaces trimmed), otherwise a potential match between a lemma and the line that contains it could be derailed by some ignorable whitespace that's carried over from the XML, or is a byproduct of the XSLT transformation.

Further, we have to make sure that only some of the text which is actually in the line is rendered; for instance, the explanations of abbreviations must be suppressed, as must line numbers.

This morning, I've expanded the JavaScript considerably, to provide normalization functions. I've also refined the routine that renders the line HTML to text (getElementText()), so that you can supply it with an argument consisting of an array of classnames to be ignored. That enables us to extract only the bits of text we want, and ignore, for instance, explanations of abbreviations. Finally, I've tweaked the XSLT so that lines with an empty lemma (where, say, the line doesn't exist in the base text) are not made into links at all.

July 6, 2007

Apparatus rendering progress

Posted by on 06 Jul 2007 in Activity log, Tasks

Worked on the rendering of the apparatus, specifically on the showing of the relevant line in a popup, with the lemma highlighted:

  • Modified poem.xsl so that it makes the lemma into a link which calls a JavaScript function.
  • Wrote JavaScript to find the relevant line.
  • Wrote functions to render out the text of the line, ready for highlighting.
  • Added a function to highlight any instances of the lemma in the line, then pop up the rendered line on the right of the screen.
  • Added CSS as appropriate (will need some tweaking for subtlety).
  • Tested.

Everything works, except that:

  • Sometimes there's a lemma which is not quite the same as the line, or portion of it. See LDev005 line 17 for an example. In this case, nothing is highlighted. That might be OK.
  • Sometimes there are elements in the line which shouldn't be part of the popup, such as notes or figures. These need to be suppressed. See LDev005 line 31 for an example.

Adding this as a task so that I get the above two things looked at before my vacation.

June 29, 2007

Fixed Apparatus bug

Posted by on 29 Jun 2007 in Activity log

Fixed a bug in Apparatus reported by Cara yesterday -- a line number tag without any lemma or reading following it would cause a list-index-out-of-bounds error. Added a bit of extra error reporting too, so that it now shows the line where the error occurred in the error message itself. Rebuilt and uploaded a new zip file.