After uploading from 2.03-dev-3 to 2.04-dev-1, whenever I try to use my
mod_perl2 applications I get a fatal error. This gets logged:

Apache::Request::config: cannot disable uploads (parser not found) at /usr/lib/perl5/Apache2/Apache/Request.pm line 64.
[Tue Apr 05 16:51:37 2005] [error] [client 132.248.72.73] Can't locate auto/Apache/Request/upload.al in @INC (@INC contains: /home/gwolf/cvs/comas_debconf/perl/Apache2 /usr/lib/perl5/Apache2 /home/gwolf/cvs/comas_debconf/perl /etc/perl /usr/local/lib/perl/5.8.4 /usr/local/share/perl/5.8.4 /usr/lib/perl5 /usr/share/perl5 /usr/lib/perl/5.8 /usr/share/perl/5.8 /usr/local/lib/site_perl .  /etc/apache2) at /home/gwolf/cvs/comas_debconf/perl/Apache2/Comas.pm line 340\n

Effectively, upload.al is not shipped with the module (neither is any
other file called *.al). I did note, though, that while the previous
version didn't ship any *upload* files, this one does:

gwolf@mosca:/tmp/libapreq2-perl-2.04-dev$ find . -name *uplo*
./docs/html/group__apreq__xs__upload.html
./docs/man/man3/apreq_xs_upload.3
./glue/perl/t/apreq/upload.t
./glue/perl/t/response/TestApReq/upload.pm

Thank you very much,

Hm, I tested the package before uploading it. Is there any special code
trigging this, or will just "use Apache::Request" do?

/* Steinar */

severity 303304 important
thanks

Hmmm... It seems this is a composite bug :-/ I am downgrading it to
severity: important, as it is not a _complete_ showstopper.

I am attaching here Apreqtest.pm with all the functionality I use,
this module _does_ work, although it generates this warning:

Apache::Request::config: cannot disable uploads (parser not found) at /usr/lib/perl5/Apache2/Apache/Request.pm line 64.
[Wed Apr 06 13:01:02 2005] [error] [client 132.248.72.73] File does not exist: /var/www/favicon.ico

This seems to be triggered by the sole presence of the DISABLE_UPLOADS
argument to the constructor of Apache2::Request - The message does no
longer appear if I comment the object creation and use only
$request=Apache::Request->new($r);

However, my system breaks with something related to this. I'm also
attaching Comas.pm, my system's entry module - It is not worth much
without the whole system, though, but you can see what I
do. Basically:

- Apache2::Comas::handler has the same logic you will find in
  Apache2::Apreqtest, anything else in it goes to the system
  initialization.
- The error happens in the _process_upload function, precisely at the
  first mention of $req->upload, which should return an empty hashref
  if no upload was received:
    return $upload unless $req->upload;

I get this in the log:

Apache::Request::config: cannot disable uploads (parser not found) at /usr/lib/perl5/Apache2/Apache/Request.pm line 64.
[Wed Apr 06 13:07:48 2005] [error] [client 132.248.72.73] Can't locate auto/Apache/Request/upload.al in @INC (@INC contains: /home/gwolf/cvs/comas_debconf/perl/Apache2 /usr/lib/perl5/Apache2 /home/gwolf/cvs/comas_debconf/perl /etc/perl /usr/local/lib/perl/5.8.4 /usr/local/share/perl/5.8.4 /usr/lib/perl5 /usr/share/perl5 /usr/lib/perl/5.8 /usr/share/perl/5.8 /usr/local/lib/site_perl . /etc/apache2) at /home/gwolf/cvs/comas_debconf/perl/Apache2/Comas.pm line 340\n

Now, if I break the function flow by returning the empty $upload
before line $340, the system works as it should (of course, except
when it should receive an upload ;-) )

I've looked a bit into this. First, latest SVN snapshot doesn't work (ie.
gives the same warning, and then immediately segfaults :-) ), so there is no
immediately known fix. Second, I don't really understand what's happening
here; it looks like the "parser" in question is something that's supposed to
be derived from the "content type" of the "environment"; it _might_ be that
the idea here is to have separate parsers depending on whatever the user
sends in. (In that case, it's logical that it can't find a parser on a GET,
since it doesn't contain a payload.)

Yes, I can reproduce it with DISABLE_UPLOADS set.

To be honest, I'm not sure if this is related or not. The documentation seems
to demand that you "use Apache::Upload" at the top, though, which seems to
fix the error for me (but the warning's still there).

/* Steinar */

Steinar H. Gunderson dijo [Wed, Apr 06, 2005 at 10:40:02PM +0200]:

Hmmm... Strangely, that rings another bug on me - a bug I didn't
report as I was unable pin it down and reproduce it - When parsing the
parameters, some of the fields were lost when receiving them via POST,
but it worked fine using POST (it was with this same script I sent
you).

Ok, I fixed this (for me) by removing the DISABLE_UPLOADS=>0, as it is
something close to a NOOP :) But... Well, the bug exists :(

Uhm, no s/POST/GET/ at some point in the middle there? :-)

To me, it sounds like the best thing to do ATM would be taking the bug
upstream; it might of course be a bug in my packaging, but it'd have to be
rather obscure, I believe :-)

/* Steinar */

Steinar H. Gunderson dijo [Thu, Apr 07, 2005 at 12:05:31AM +0200]:

http://comas.linux-aktivaattori.org/debconf5/committee/academic_committee/41?-posted=1&-title=Debian-Kernel+Team+Oveview+and+Status&-abstract=I+would+like+to+lead+attending+members+of+the+debian-kernel+team+in+a+talk+that+will+serve+as+a+status+update+for+our+team.++The+Debian+Kernel+Team+has+been+around+for+about+a+year+now.%0D%0A%0D%0ADiscussion+topics+would+be+mostly+technical%2C+but+would+include+both+social+%28how+we+work+together%29+and+political+%28non-free+firmware%29+as+well.%0D%0A%0D%0ATechnical+topics+include%3A%0D%0AIntroduction+-+Who+are+we%2C+and+what+do+we+do%3F%0D%0A++++++*+Maintenance+of+kernel-%5Bsource%2Cimage%2Cpatch-ARCH%5D%2C+initrd-tools%2C+etc%0D%0A++++++++++++++*+security+for+testing%2Funstable%0D%0A++++++++++++++*+Package+architecture.++What+packages+does+our+team+create%2C+and+what+are+they+for%3F++%0D%0A++++++*+kernel-tree+vs.+kernel-source%0D%0A++++++++++++++*+What+kernel-headers+do+I+need%3F%0D%0A++++++++++++++*+What+is+the+right+way+to+file+a+bug%3F%0D%0A++++++*+What+bugs+should+be+filed+against+the+%27kernel%27+package%2C+vs.+a+real+package.%0D%0A++++++++++++++*+Classes+of+bugs%0D%0A++++++++++++++*+Build+issues%0D%0A++++++++++++++++++++++*+Runtime+issues%0D%0A++++++++++++++++++++++*+Where+can+you+find+our+source%3F++An+overview+of+our+svn+repo.%0D%0A++++++*+Daily+build+testing+-+Simon+Horman+and+I+have+both+done+some+work+in+this+area%3B+hopefully+we%27ll+have+some+results+by+DebConf5+time.%0D%0A++++++*+What+requirements+must+a+patch+fulfill+before+we+will+include+it%0D%0A++++++*+Interacting+with+the+debian-installer+team+%28l-k-di%29%0D%0A++++++*+ABI+changes%3B+what+they%27re+for+and+why+the+suck%0D%0A++++++*+Exist+for+module+compatability%0D%0A++++++++++++++*+Break+upgrades+%28anecdotal+ia64%2Fvfat+issue%3F%29%0D%0A++++++++++++++*+How+ABI+breaks+break+debian-installer%0D%0A++++++++++++++*+Testing+for+ABI+breaks%0D%0A++++++++++++++*+How+to+build+additional+modules+for+our+precompiled+kernel-images%0D%0A++++++*+How+the+architectures+work+together+%28or+not%29+-+sharing+configs%2C+build+system%3B+sharing+source+vs.+kernel-patch-ARCH+packages.%0D%0A++++++*+All-kernels-from-one-source-package+plans.++Sven+Luther+has+been+looking+into+building+most+%28or+all%29+of+our+kernels+from+a+single+source+package.++What+are+the+benefits%3F++What+are+the+bottlenecks%3F++What+is+the+status%3F%0D%0A++++++*+What+areas+can+we+use+help%3F%0D%0A++++++*+Firmware%3B+what+is+acceptable+and+what%27s+not.++This+topic+maybe+one+to+avoid%2C+due+to+its+political+nature.++It+maybe+better+to+fork+this+off+into+a+BOF.%0D%0A++++++*+%0D%0AWhy+am+I+qualified+for+this+talk%3F++The+principle+reason+is+that+I+won%27t+be+developing+the+content+or+speaking+alone.++Topics+are+being+worked+out+with+the+debian-kernel+team%2C+and+there+will+be+at+least+a+few+of+us+around+to+give+the+actual+talk.++I+began+a+discussion+of+topics+last+month%3A%0D%0A++http%3A%2F%5D2F2F2Flists.debian.org%2Fdebian-kernel%2F2005%2F02%2Fmsg00017.html%0D%0A%0D%0AI+also+plan+to+prepare+the+talk+with+the+group.++I%27ve+setup+a+wiki+for+this+purpose%3A%0D%0A++http%3A%2F%2Fwiki.debian.net%2F%3FDebianKernelDebconf5%0D%0A%0D%0A%0D%0AMy+personal+involvement+with+the+debian-kernel+team+has+been+mostly+as+a+port+maintainer.++I+took+over+maintenance+of+the+ia64+kernel+packages+from+Bdale+a+little+over+a+year+ago+now%2C+and+I+co-maintained+these+packages+for+some+time+before+that.++I+converted+the+packages+from+a+rather+manual+system+to+an+architecture+similar+to+Herbert+Xu%27s%2C+and+began+tracking+his+kernel-source+packages.++Once+I+joined+the+kernel+team%2C+I+worked+with+Christopher+Hellwig+to+merge+the+ia64+2.6+patches+into+our+shared+kernel-source+package.++I+also+created+the+ia64+versions+of+the+linux-kernel-di+packages.++Outside+of+my+porter+role%2C+I+help+out+with+the+occasional+security+patch+or+bug+fix.%0D%0A%0D%0AQualification+Links%3A%0D%0A+The+2.6%2Fia64+subtree+of+our+svn+repository%3A%0D%0A++http%3A%2F%2Fsvn.debian.org%2Fwsvn%2Fkernel%2Ftrunk%2Fkernel%2Fia64%2F%3Frev%3D0%26sc%3D0%0D%0A+A+couple+of+tools+I+created+to+provide+daily+status%3A%0D%0A++http%3A%2F%2Fpeople.debian.org%2F%7Edannf%2Fkernel-stats%2Fkernel-avail.html%0D%0A++http%3A%2F%2Fpeople.debian.org%2F%7Edannf%2Fkernel-stats%2Fkern-dep.html%0D%0ASome+documentation+I+wrote%3A%0D%0A++http%3A%2F%2Fwiki.debian.net%2F%3FDebianKernelTree%0D%0A+My+first+archived+commit%3A%0D%0Ahttp%3A%2F%2Flists.alioth.debian.org%2Fpipermail%2Fkernel-svn-changes%2F2004-July%2F000021.html%0D%0AAnd+participation+in+what+I+think+was+our+only+formal+irc+meeting%3A%0D%0A++http%3A%2F%2Fminbar.dodds.net%2F%7Evorlon%2Fkernel-2.6.10-discussion.log%0D%0A&-track_id=2&-prop_type_id=1&-prop_status_id=1&-comments=&-my_rating=4&-my_internal_comment=test

is equivalent to this POST request: (only relevant fields copied)

POST /debconf5/committee/academic_committee/41 HTTP/1.1
Host: comas.linux-aktivaattori.org
Content-Length: 4653
-abstract=I+would+like+to+lead+attending+members+of+the+debian-kernel+team+in+a+talk+that+will+serve+as+a+status+update+for+our+team.++The+Debian+Kernel+Team+has+been+around+for+about+a+year+now.%0D%0A%0D%0ADiscussion+topics+would+be+mostly+technical%2C+but+would+include+both+social+%28how+we+work+together%29+and+political+%28non-free+firmware%29+as+well.%0D%0A%0D%0ATechnical+topics+include%3A%0D%0AIntroduction+-+Who+are+we%2C+and+what+do+we+do%3F%0D%0A++++++*+Maintenance+of+kernel-%5Bsource%2Cimage%2Cpatch-ARCH%5D%2C+initrd-tools%2C+etc%0D%0A++++++++++++++*+security+for+testing%2Funstable%0D%0A++++++++++++++*+Package+architecture.++What+packages+does+our+team+create%2C+and+what+are+they+for%3F++%0D%0A++++++*+kernel-tree+vs.+kernel-source%0D%0A++++++++++++++*+What+kernel-headers+do+I+need%3F%0D%0A++++++++++++++*+What+is+the+right+way+to+file+a+bug%3F%0D%0A++++++*+What+bugs+should+be+filed+against+the+%27kernel%27+package%2C+vs.+a+real+package.%0D%0A++++++++++++++*+Classes+of+bugs%0D%0A++++++++++++++*+Build+issues%0D%0A++++++++++++++++++++++*+Runtime+issues%0D%0A++++++++++++++++++++++*+Where+can+you+find+our+source%3F++An+overview+of+our+svn+repo.%0D%0A++++++*+Daily+build+testing+-+Simon+Horman+and+I+have+both+done+some+work+in+this+area%3B+hopefully+we%27ll+have+some+results+by+DebConf5+time.%0D%0A++++++*+What+requirements+must+a+patch+fulfill+before+we+will+include+it%0D%0A++++++*+Interacting+with+the+debian-installer+team+%28l-k-di%29%0D%0A++++++*+ABI+changes%3B+what+they%27re+for+and+why+the+suck%0D%0A++++++*+Exist+for+module+compatability%0D%0A++++++++++++++*+Break+upgrades+%28anecdotal+ia64%2Fvfat+issue%3F%29%0D%0A++++++++++++++*+How+ABI+breaks+break+debian-installer%0D%0A++++++++++++++*+Testing+for+ABI+breaks%0D%0A++++++++++++++*+How+to+build+additional+modules+for+our+precompiled+kernel-images%0D%0A++++++*+How+the+architectures+work+together+%28or+not%29+-+sharing+configs%2C+build+system%3B+sharing+source+vs.+kernel-patch-ARCH+packages.%0D%0A++++++*+All-kernels-from-one-source-package+plans.++Sven+Luther+has+been+looking+into+building+most+%28or+all%29+of+our+kernels+from+a+single+source+package.++What+are+the+benefits%3F++What+are+the+bottlenecks%3F++What+is+the+status%3F%0D%0A++++++*+What+areas+can+we+use+help%3F%0D%0A++++++*+Firmware%3B+what+is+acceptable+and+what%27s+not.++This+topic+maybe+one+to+avoid%2C+due+to+its+political+nature.++It+maybe+better+to+fork+this+off+into+a+BOF.%0D%0A++++++*+%0D%0AWhy+am+I+qualified+for+this+talk%3F++The+principle+reason+is+that+I+won%27t+be+developing+the+content+or+speaking+alone.++Topics+are+being+worked+out+with+the+debian-kernel+team%2C+and+there+will+be+at+least+a+few+of+us+around+to+give+the+actual+talk.++I+began+a+discussion+of+topics+last+month%3A%0D%0A++http%3A%2F%5D2F2F2Flists.debian.org%2Fdebian-kernel%2F2005%2F02%2Fmsg00017.html%0D%0A%0D%0AI+also+plan+to+prepare+the+talk+with+the+group.++I%27ve+setup+a+wiki+for+this+purpose%3A%0D%0A++http%3A%2F%2Fwiki.debian.net%2F%3FDebianKernelDebconf5%0D%0A%0D%0A%0D%0AMy+personal+involvement+with+the+debian-kernel+team+has+been+mostly+as+a+port+maintainer.++I+took+over+maintenance+of+the+ia64+kernel+packages+from+Bdale+a+little+over+a+year+ago+now%2C+and+I+co-maintained+these+packages+for+some+time+before+that.++I+converted+the+packages+from+a+rather+manual+system+to+an+architecture+similar+to+Herbert+Xu%27s%2C+and+began+tracking+his+kernel-source+packages.++Once+I+joined+the+kernel+team%2C+I+worked+with+Christopher+Hellwig+to+merge+the+ia64+2.6+patches+into+our+shared+kernel-source+package.++I+also+created+the+ia64+versions+of+the+linux-kernel-di+packages.++Outside+of+my+porter+role%2C+I+help+out+with+the+occasional+security+patch+or+bug+fix.%0D%0A%0D%0AQualification+Links%3A%0D%0A+The+2.6%2Fia64+subtree+of+our+svn+repository%3A%0D%0A++http%3A%2F%2Fsvn.debian.org%2Fwsvn%2Fkernel%2Ftrunk%2Fkernel%2Fia64%2F%3Frev%3D0%26sc%3D0%0D%0A+A+couple+of+tools+I+created+to+provide+daily+status%3A%0D%0A++http%3A%2F%2Fpeople.debian.org%2F%7Edannf%2Fkernel-stats%2Fkernel-avail.html%0D%0A++http%3A%2F%2Fpeople.debian.org%2F%7Edannf%2Fkernel-stats%2Fkern-dep.html%0D%0ASome+documentation+I+wrote%3A%0D%0A++http%3A%2F%2Fwiki.debian.net%2F%3FDebianKernelTree%0D%0A+My+first+archived+commit%3A%0D%0Ahttp%3A%2F%2Flists.alioth.debian.org%2Fpipermail%2Fkernel-svn-changes%2F2004-July%2F000021.html%0D%0AAnd+participation+in+what+I+think+was+our+only+formal+irc+meeting%3A%0D%0A++http%3A%2F%2Fminbar.dodds.net%2F%7Evorlon%2Fkernel-2.6.10-discussion.log%0D%0A&-track_id=2&-prop_type_id=1&-prop_status_id=1&-comments=&-my_internal_comment=test&-posted=1&-title=Debian-Kernel+Team+Oveview+and+Status&-my_rating=4

But (sometimes) the result is not the same. This time it worked... But
I cannot always reproduce it :-/

Agree

Whoa. :-)

OK, I'll leave it open; let me know how your conversations with upstream work
out :-)

/* Steinar */