Last modified: 2011-02-23 18:19:41 UTC
PageA contains: "This is a wo" PageB contains: "rd between pages" Transcluding those pages using <pages> tag results in a text: "This is a wo rd between pages", but expected text is: "This is a word between pages".
this is expected behaviour
This _was_ expected behaviour? Older versions did not include that whitespace, so on pl.wikisource pages contain exact text, that appears on the scans - it makes sense. Now divided words have space in the middle, so we need to review our all pages, and correct that moving word to one of the pages. It is counterintuitive...
this is very strange. are you sure ?
I am sure, because we have lots of pages, which have divided words. They were always glued together by <pages> without whitespaces. If someone wanted to change that behaviour, it could have been done simply by adding whitespace to [[MediaWiki:Proofreadpage pagenum template]].
Yes, we are completely sure, older versions did not inlcuded whitespaces. We have over 20 000 pages with exact text, down to the spaces. Examples: http://pl.wikisource.org/wiki/Strona:PL_Gloger-Encyklopedja_staropolska_ilustrowana_T.2_123.jpg and http://pl.wikisource.org/wiki/Strona:PL_Gloger-Encyklopedja_staropolska_ilustrowana_T.2_124.jpg As you can see, whitespaces were definitely not expected on those pages.
I just dicussed with ankry on IRC. Apparently there was a bug in the software, that caused whitespace not to be added when pages are numbered manually. I was not aware of it. On other wikisources, manual numbering of pages is almost never used in conjunction with <pages/>, so this bug went unnoticed. I do not think that we should revert to the former behaviour; it does not make sense to keep incompatibilities. in addition, it is cumbersome to add   at the end of all pages. pl.ws is a small wiki. we can fix it with a robot.
I would like to avoid using a bot. Maybe there can be introduced configuration variable, which allows to switch to older behaviour? Or if it went unnoticed, just leave the old behaviour and add whitespace to [[MediaWiki:Proofreadpage pagenum template]] on projects where it is needed.
[[MediaWiki:Proofreadpage pagenum template]] is for page numbers and it might be deprecated in the future; better don't use it for something else.
I hate writing ugly pieces of code, but if there is no other way, we will just use javascript to remove that whitespace for now. It won't work for every user, but I think it is better to keep exact text on pages without using additional templates or other wikicode solutions.
if the absence of   currently predicts the need for a hyphenation template, you should not remove it ; it can be used by a robot to add the template where it is needed.