On Tue, 2011-09-20 at 16:21:27 -0700, Kaz Kylheku wrote:
On Tue, 20 Sep 2011 19:09:15 -0400, Sahil Tandon sahil@tandon.net wrote:
Although the INSTALL document recommends the inclusion of "text/css" in Lighty's compress.filetype array, removing "text/css" (i.e. not compressing css) restores normal UI behavior. I am not entirely satisfied with this, but it is a sufficient workaround for now.
Good work. Okay, so the problem would seem to be in the compression or decompression. Might it be a issue in lighthttpd?
That would be the obvious conclusion but such diagnosis is too general and, thus, incomplete. For example, other css-employing web sites served by the same lighttpd instance are unaffected by css compression.
You're missing just perhaps one more step in your excellent root-cause analysis.
I think more than one step is missing. Indeed, that is why I am on this seeking help from the experts.
Namely, google for "lighthttpd broken compression".
Been there; done that.
First hit:
Yes, I noticed this when I conducted similar google queries several days ago.
"Lighthttpd - css files broken if mod_compress enabled on lighttpd 1.4.26"
A careful reading of that link will reveal nuanced differences between the issue reported by "telix" and my problem.
The OP in that thread is apparently referring to broken css files even when text/css is not explicitly set in compress.filetype. What I have reported above is that RC behaves as desired when text/css is not set in compress.filetype, but I did not have to disable mod_compress! In my testing, it is *only* when compress.filetype is configured to compress text/css does the problem manifest itself in various browsers. Compressing other mimetypes is just fine, does not cause any problems, and there is no apparent breakage of css or RC.
So, I would appreciate if others who are using lighttpd + roundcube 0.5.4 (or more recent trunk) *and* compressing text/css mimetype would share their lighttpd configurations as they related to RC.
In the meantime, I will continue to investigate for "root causes" and report my findings.
List info: http://lists.roundcube.net/users/ BT/9b404e9e