Last modified: 2014-07-28 15:25:00 UTC

Wikimedia Bugzilla is closed!

Wikimedia migrated from Bugzilla to Phabricator. Bug reports are handled in Wikimedia Phabricator.
This static website is read-only and for historical purposes. It is not possible to log in and except for displaying bug reports and their history, links might be broken. See T69928, the corresponding Phabricator task for complete and up-to-date bug report information.
Bug 67928 - Vagrant: double-gzipped styles output to Safari
Vagrant: double-gzipped styles output to Safari
Status: RESOLVED FIXED
Product: MediaWiki-Vagrant
Classification: Unclassified
General (Other open bugs)
unspecified
All All
: Unprioritized normal
: ---
Assigned To: Ori Livneh
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2014-07-12 16:44 UTC by Brion Vibber
Modified: 2014-07-28 15:25 UTC (History)
3 users (show)

See Also:
Web browser: ---
Mobile Platform: ---
Assignee Huggle Beta Tester: ---


Attachments
Screenshot of Safari network debugger panel for broken styles (161.92 KB, image/png)
2014-07-12 16:44 UTC, Brion Vibber
Details

Description Brion Vibber 2014-07-12 16:44:33 UTC
Created attachment 15923 [details]
Screenshot of Safari network debugger panel for broken styles

The styles as loaded from my freshly installed mediawiki-vagrant:

http://localhost:8080/w/load.php?debug=false&lang=en&modules=mediawiki.legacy.commonPrint%2Cshared%7Cmediawiki.skinning.interface%7Cmediawiki.ui.button%7Cskins.vector.styles&only=styles&skin=vector&*

somehow ends up delivering raw gzipped data in Safari, while I see the expected decompressed CSS text in Chrome and Firefox. This of course means an unstyled site which is not very usable. :D

I can actually save the garbled file to disk and gunzip it from there...

Doesn't seem to be browser cache, I can repro it on a second machine or by changing URL.

Sometimes clears up if I manually remove or rearrange some modules from the query string, but other times it gives me a different, but still extra-zipped, data set.

Safari's debugger indicates the file is marked as gzipped, so it seems to be double-gzipped rather than missing the header.


Interestingly, Content-Encoding header says "gzip, gzip". Hmmm.
Comment 1 Brion Vibber 2014-07-12 16:53:11 UTC
Chrome and Firefox show the same "Content-Encoding: gzip, gzip" ... I think Chrome & Firefox are actually applying decompression twice, while Safari is not.

Is something intermittently adding a second layer of gripping?

HTTP 1.1 spec seems to indicate that multiple content-encodings are legit, but I wouldn't recommend relying on it. :)
Comment 2 Gerrit Notification Bot 2014-07-12 17:00:33 UTC
Change 145771 had a related patch set uploaded by Ori.livneh:
hhvm: set GzipCompressionLevel = 0 to avoid double-gzip

https://gerrit.wikimedia.org/r/145771
Comment 3 Gerrit Notification Bot 2014-07-12 17:06:42 UTC
Change 145771 merged by jenkins-bot:
hhvm: set GzipCompressionLevel = 0 to avoid double-gzip

https://gerrit.wikimedia.org/r/145771
Comment 4 Brion Vibber 2014-07-12 17:10:03 UTC
Confirmed that ori's patch fixes the double gzipping for me. Yay!
Comment 5 Antoine "hashar" Musso (WMF) 2014-07-28 15:25:00 UTC
The same double gzip occurs on beta bug 68720

Note You need to log in before you can comment on or make changes to this bug.


Navigation
Links