Last words before tagging release

["Gregory (Grisha) Trubetskoy" <gr...@apache.org>]

I think that the diminished activity on bug reporting on this and the
other python lists indicates that the latest cvs could be tagged and
released as 3.1.3 stable.

So if you have any objections, speak up now.

Grisha

Re: Last words before tagging release

[Ron Reisor <ro...@UDel.Edu>]

The one patch I still need to add to the cvs to get it to work on MacOSX
is this patch. I believe it's contributed by Mack McClain <ma...@gsoa.net>.
Can this be added to the release?

*** httpd-python/configure	Mon Feb  9 12:39:02 2004
--- httpd-python-udel/configure	Mon Dec 15 14:24:35 2003
***************
*** 1299,1304 ****
--- 1299,1305 ----
  if test "$PyFRAMEWORKDIR" != "no-framework"; then
    if test -n "$PyFRAMEWORK"; then
      PyPYTHONLIBS="-framework $PyFRAMEWORK"
+     LDFLAGS="${LDFLAGS} -Wl,-framework,Python"
    fi
  else
    LDFLAGS="${LDFLAGS} -L${PyLIBPL}"

cheers,

Ron

Ron Reisor <ro...@udel.edu> (RWR3)
University of Delaware Information Technologies/Network and Systems Services
Computing Center/192 South Chapel Street/Newark DE, 19716
pgp finger print: 0D 73 06 6F D3 6A 99 D3  F5 D5 6E FF 3B B9 7C 2C

Build problems on FreeBSD with flex

[Barry Pederson <bp...@barryp.org>]

Gregory (Grisha) Trubetskoy wrote:
> I think that the diminished activity on bug reporting on this and the
> other python lists indicates that the latest cvs could be tagged and
> released as 3.1.3 stable.
> 
> So if you have any objections, speak up now.

Here's something else...I just fetched httpd-python using CVS, followed 
the instructions in the README for building by running

    ./configure --with-apxs=/usr/local/sbin/apxs
    make

But the make dies with:

-----------
/usr/local/bin/flex -R -opsp_parser.c --header-file=include/psp_flex.h 
psp_parser.l
/usr/local/bin/flex: not found
*** Error code 127

Stop in /root/httpd-python/src.
*** Error code 1

Stop in /root/httpd-python/dist.
*** Error code 1

Stop in /root/httpd-python.
------------

Any chance of having configure locate flex for you, rather than having 
it hardcoded in?

On a FreeBSD box, flex is in /usr/bin - but if I symlink 
/usr/local/bin/flex to /usr/bin/flex, make then dies with:

-----------
/usr/local/bin/flex -R -opsp_parser.c --header-file=include/psp_flex.h 
psp_parser.l
/usr/local/bin/flex: unknown flag 'R'.  For usage, try
         /usr/local/bin/flex --help
-----------

I see a comment in the httpd-python/src/Makefile

    # requires flex 2.5.31 for reentrant support

but FreeBSD 5.2 and 4.8 only seem to have 2.5.4, and I don't see
any ports that offer a newer version.  So it doesn't seem like 
mod_python 3.1.x is going to play well with the FreeBSD ports system at 
the moment.  I'd imagine the README would have to tell FreeBSD users to 
build and install a newer flex by hand unless there's a way to get 
around needing reentrancy :(

How about a configure option to skip support for psp pages (and the need 
for flex), so we could do something like "./configure --disable-psp" if 
need be?

	Barry

Re: Last words before tagging release

[Barry Pederson <bp...@barryp.org>]

Gregory (Grisha) Trubetskoy wrote:
> I think that the diminished activity on bug reporting on this and the
> other python lists indicates that the latest cvs could be tagged and
> released as 3.1.3 stable.
> 
> So if you have any objections, speak up now.

This is a very vague complaint, and I haven't had time to really look 
into it, but memory leaks seem to be a problem at least on my machines 
(FreeBSD 4.8 and 5.2, Apache 2.0.48 preforked, Python 2.3.3)- both with 
3.1.x and 3.0.x.  In some cases I seem to be losing >100k per request. 
Calling Python's gc.collect() doesn't help, so I'd suspect something 
with mod_python and apache.

Any suggestions as to how to go about tracking something like that down?
I've fooled around with running httpd with the "-X" debugging option, 
to force all the requests to go through a single process - but I haven't 
figured out a way to look inside and see where the memory went.

I know that's not much for other people to go on, but has anyone else 
done testing for this sort of thing?  Even a "hello world" type app 
requested a few thousand times?

	Barry