https://bz.apache.org/bugzilla/show_bug.cgi?id=69575
Christopher Schultz <ch...@christopherschultz.net> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |REOPENED
Resolution|FIXED |---
--- Comment #3 from Christopher Schultz <ch...@christopherschultz.net> ---
Unfortunately, there is not really a generalizable solution for
double-compression, but I think we /can/ do better.
Some Content-Encoding values indicate that compression is being performed (e.g.
gzip, deflate, lzma, br, etc.) while others mean ... something else entirely
(e.g. exi, rsync, and some odd Microsoft-defined things). Of course there is
also "identity" which we can understand to be trivially ignorable.
Most of the tokens that we would consider "problematic" would be the ones that
are far less common (e.g. Microsoft stuff, rsync, etc.) and maybe we can ignore
those.
Skipping gzip compression on known-compressed Content-Encodings (gzip, deflate,
br, compress, and zstd) is definitely worth implementing as Rémy has done here,
so that's good.
It is far more "correct" to append ", gzip" to any Content-Encoding that is
already present (unless it's exactly "identity"), though Tomcat tends to be an
origin server and not a proxy. I think we should append and not replace.
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
