Page MenuHomePhabricator

Missing whitespaces between words in article's last line on hy.wp
Open, MediumPublic

Description

Author: norayr

Description:
lost whitespaces in the last line.

Whitespaces between last words of the page disappear sometimes. I have noticed the issue first in my own mediawiki's but today eventually have seen it online on Wikipedia.

This is the article for example: http://hy.wikipedia.org/wiki/%D5%87%D5%A1%D5%BC%D5%AC_%D4%B9%D5%A5%D5%B8%D5%A4%D5%B8%D6%80_%D4%B1%D5%B6%D6%80%D5%AB_%D5%A4%D5%A8_%D4%BF%D5%B8%D5%BD%D5%BF%D5%A5%D6%80

If one moves Categories to the upper part of the text, the problem disappears. I have noticed it since, I guess, 17th version of mediawiki.

Thank you


Version: 1.23.0
Severity: normal

Attached:

hy.wikipedia.org_2014-04-15_02-10-41.png (1×1 px, 270 KB)

Details

Reference
bz63918

Event Timeline

bzimport raised the priority of this task from to High.Nov 22 2014, 3:23 AM
bzimport set Reference to bz63918.

Thanks for taking the time to report this!

Confirming that the text should be
Մահացել է 1879 թվականի մայիսի 7-ին, մենակ և կարիքավոր վիճակում։
but the last space is missing, using Firefox 28 on Fedora 20.

Can still be seen on https://hy.wikipedia.org/wiki/Շառլ_Թեոդոր_Անրի_դը_Կոստեր

Մահացել է 1879 թվականի մայիսի 7-ին, մենակ և կարիքավոր վիճակում։

can be seen in the wikitext when editing the page.

But the HTML source of the page says:
<p>Մահացել է 1879 թվականի մայիսի 7-ին, մենակ ևկարիքավորվիճակում։</p>

I can also confirm that removing this section fixes the problem:
{{DEFAULTSORT:դը Կոստեր, Շառլ}}
[[Կատեգորիա:Բելգիացի գրողներ]]
[[Կատեգորիա:1827 ծնունդներ]]
[[Կատեգորիա:Օգոստոսի 20 ծնունդներ]]
[[Կատեգորիա:1879 մահեր]]
[[Կատեգորիա:Մայիսի 7 մահեր]]

Gabriel, James: As this seems to be Parser territory, any idea who could investigate here? See summary and testcase in comment 2.

Adding Subbu as parser/Parsoid leader.

Confirmed that this is PHP parser only. http://parsoid-lb.eqiad.wikimedia.org/hywiki/%D5%87%D5%A1%D5%BC%D5%AC_%D5%A4%D5%A8_%D4%BF%D5%B8%D5%BD%D5%BF%D5%A5%D6%80?oldid=1468063 shows the right output from Parsoid.

This bug smells a bit like a bad regexp because of the presence of some characters.

Scott has traditionally poked around the PHP parser but he is currently occupied with PDF migration. We are currently stretched a bit thin, but we'll reassign to one of us if this is still not addressed in next 3 weeks (we have our quarterly review coming up). But, dont want to prematurely assign it to one of us if someone else can investigate.

(In reply to ssastry from comment #6)

we'll reassign to one of us if this is still not addressed in next 3 weeks

4 weeks are over. ssastry?

I can probably take a look at this, but it's not high priority.

Can someone reproduce this on enwiki? It would make writing a test case much easier if so.

(As a first step, copy-and-pasting the wikitext from hywiki to a sandbox on enwiki and looking at the render would let us know if the bug relates to a hywiki setting (ie, link prefix or link suffix).)

I've made English only text, to check if this has something to do with page content language.

This doesn't reproduce on en:WP (https://en.wikipedia.org/wiki/User:Xelgen/sandbox)
This also doesn't reproduce on en:WP when you switch to Armenian UI language.

Same time issue is being observed on hy:WP with English UI language. (http://hy.wikipedia.org/wiki/%D5%84%D5%A1%D5%BD%D5%B6%D5%A1%D5%AF%D5%AB%D6%81:Xelgen/LastWhitespace)

But I doubt it's solely WMF configuration issues, as in first comment report mentioned he observed this on his own MediaWiki installation.

For example:
http://chtesutyun.arnet.am/index.php/%D5%84%D5%A1%D5%BD%D5%B6%D5%A1%D5%AF%D5%AB%D6%81:Xelgen
http://grapaharan.org/index.php?title=%D5%84%D5%A1%D5%BD%D5%B6%D5%A1%D5%AF%D5%AB%D6%81:Xelgen/LastWhitespace

Aklapper lowered the priority of this task from High to Medium.Apr 17 2015, 1:44 PM

Would be awesome if someone find time to track this down, but looks like realistically speaking this is not high priority currently. :(