Patches pending commit for 2.3
Felix Schwarz
today I committed a few trivial patches to the 2.3.x branch. Now I glanced
over the bugs for newer versions to find out if there is more stuff that needs
backporting.
This is my preliminary queue of pending patches, feel free to comment/discuss:
- #995 - CherryPy 2.3.0 regression with Python 2.5
http://www.bitbucket.org/felixschwarz/cp2-candidates/src/tip/invalid_cookies
Even though Python changed it's behavior for Python 2.5+, I don't see any
actual change in the tests so I have to commit that without any test case...
Interestingly this is fixed differently in 3.1 and trunk. Also 3.0 misses
the fix completely. To me the fix in 3.1 (r1765) looks a bit better so I'm
tempted to backport this instead of the patch in #995.
Also #735 which is the original issue is still open. I think it should be
closed again as it changed scope (it was fixed but no in the version the
last commenter wanted)
- #868 - traceback returned when CherryPy receives invalid cookie
http://www.bitbucket.org/felixschwarz/cp2-candidates/src/tip/invalid_cookies
backport of r2175
- #951 - Content-Length breaks curl on 100-Continue
no patch yet but looks interesting enough to warant a backport.
Debian also ships a patch for the autoreloader:
http://patch-tracker.debian.org/package/python-cherrypy/2.3.0-3
Basically it's a backport of later changes to _cpengine so I guess that can go
in as well...
fs
Christoph Zwerschke
> Interestingly this is fixed differently in 3.1 and trunk. Also 3.0 misses
> the fix completely. To me the fix in 3.1 (r1765) looks a bit better so I'm
> tempted to backport this instead of the patch in #995.
Do you mean http://www.cherrypy.org/changeset/1765? To me that's the
uglies fix of all. Either output cookie with sep="\n" and then split
with "\n", or output cookie with no (i.e. default) sep, and then split
with splitlines() - the splitlines() handles all kinds of line seps. The
latter method looks like the cleanest solution to me, plus it also
obviates the need for the "if cookie" statement since the list will be
empty in this case.
> Also #735 which is the original issue is still open. I think it
> should be closed again as it changed scope (it was fixed but no in
> the version the last commenter wanted)
Right, that's the same issue and can be closed then.
From my side, I don't have any more patches for 2.3.1; I noticed the
deprecation of sha and md5 in Py 2.6 was already handled in r2662.
-- Christoph
Robert Brewer
> today I committed a few trivial patches to the 2.3.x branch.
> Now I glanced over the bugs for newer versions to find out
> if there is more stuff that needs backporting.
>
> This is my preliminary queue of pending patches, feel free to
> comment/discuss:
> - #995 - CherryPy 2.3.0 regression with Python 2.5
> http://www.bitbucket.org/felixschwarz/cp2-
> candidates/src/tip/invalid_cookies
> Even though Python changed it's behavior for Python 2.5+, I don't
> see any
> actual change in the tests so I have to commit that without any test
> case...
>
> Interestingly this is fixed differently in 3.1 and trunk. Also 3.0
> misses the fix completely.
> To me the fix in 3.1 (r1765) looks a bit better
> so I'm
> tempted to backport this instead of the patch in #995.
>
> Also #735 which is the original issue is still open. I think it
> should be
> closed again as it changed scope (it was fixed but no in the version
> the
> last commenter wanted)
> - #868 - traceback returned when CherryPy receives invalid cookie
> http://www.bitbucket.org/felixschwarz/cp2-
> candidates/src/tip/invalid_cookies
> backport of r2175
>
> - #951 - Content-Length breaks curl on 100-Continue
> no patch yet but looks interesting enough to warant a backport.
>
> Debian also ships a patch for the autoreloader:
> http://patch-tracker.debian.org/package/python-cherrypy/2.3.0-3
> Basically it's a backport of later changes to _cpengine so I guess that
> can go
> in as well...
Robert Brewer
fuma...@aminus.org