You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Maxime Petazzoni <ma...@bulix.org> on 2005/12/12 01:45:18 UTC
A mod_mbox release ?
Hi,
Some of you must remember when, a few months ago, I came -almost lost
in Stuttgart!- to the last Hackathon in order to start hacking on
mod_mbox for my SoC project.
The ApacheCon US 2005 has finally arrived, and I think it's time to
look back to what have been done in mod_mbox. During the summer and
according to Google's SoC rules, most of the development was done in
order to get a "new" mod_mbox. Since September, a lot of bug fixes
where done and the thing is now running quite well on
mail-archives.apache.org. As of tonight, revision 356073 makes
mod_mbox compile and run with latest GCCs and APR/APR-Util/HTTPd
trunks.
Apache HTTPd 2.2 was released a dozen days ago, and I believe we
should now think about (finally) releasing mod_mbox 0.2. Of course,
there's still a lot to do on the module (check out the STATUS file for
more information), but I think we could tag current module's status as
a "Summer Of Code release" and make it available as a packaged version
(ie. not only from Svn).
I'm not yet used to release management (even for a small module like
mod_mbox), so I'll be very pleased to get some feedback, comments and
pointers if we decide to make a 0.2 release for mod_mbox !
Regards,
- Sam
--
Maxime Petazzoni (http://www.bulix.org)
-- gone crazy, back soon. leave message.
Re: A mod_mbox release ?
Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
On Sun, Dec 11, 2005 at 08:06:54PM -0800, Sander Temme wrote:
> Good idea: releasing the code in packaged form should encourage its use.
>
> The main aspect of Release Management is (all IMHO of course) making
> an informed judgement of whether the current code base is "good
> enough" to release. This includes data points like:
mod_mbox has never been formally released as we've never had three
committers interested.
I'd prefer we just take whatever we have now, try to get some feedback
(perhaps announce it on announce@httpd, but I don't know), and then see
where it goes from there. If folks contribute the finer points, cool.
Otherwise, well, it'll be packaged so that it'll slightly ease the barriers
to adoption compared to where we are now.
Anyhow, a big +1 for a release here. I will definitely test any releases
that are posted. =) -- justin
Re: A mod_mbox release ?
Posted by Justin Erenkrantz <ju...@erenkrantz.com>.
On Mon, Dec 12, 2005 at 09:38:16AM +0100, Maxime Petazzoni wrote:
> That's why I only want to call it 0.2 and not 1.0. Because a 0.3 will
> come in the next months I hope to fix these problems and improve the
> thing. I don't think we should call it 1.0 until we make the dynamic
> browser work everywhere. Google makes it for every single
> bleeding-edge web-based application they do, why not us ?
I agree that's a suitable roadmap for mod_mbox 1.0. -- justin
Re: A mod_mbox release ?
Posted by Joshua Slive <jo...@slive.ca>.
Maxime Petazzoni wrote:
>> * It's been running on mail-archives.apache.org with fewer than X
>> cores (with X tending to 0)
>
> I don't know if I'm able to check this point on my own : where do
> coredumps go ? Do I have enough access rights to check for them ?
>
> Anyway, since my last fixes against core dumps, you did not report any
> of them.
I was going to chime in saying that there were no recent cores, but then
we had one this morning ;-(. Details below. But anyway, I think things
look pretty good given the substantial volume of traffic that mod_mbox
serves, including search engines that tend to tickle every obscure
messages and links.
For the coredump, the request was
"GET
/mod_mbox/db-derby-dev/200507.mbox/%3C42EA7087.40900@Sourcery.Org%3E
HTTP/1.1"
and the full backtrace:
#0 0x2000000000594940 in strstr () from /lib/tls/libc.so.6.1
No symbol table info available.
#1 0x2000000001020630 in mbox_mime_decode_multipart (p=0x60000000004b93a8,
body=0x3ffffffffef45b97 <Address 0x3ffffffffef45b97 out of bounds>,
ct=0x60000000003d05c8 "text/plain", cte=CTE_7BIT, boundary=0x0)
at mod_mbox_mime.c:34
mail = (mbox_mime_message_t *) 0x60000000002e0f90
tmp = 0xc000000000000590 <Address 0xc000000000000590 out of bounds>
k = 0x60000000003d0000 "ยจ\223K"
end_bound = 0x200000000102b9a0 "\n\n"
#2 0x200000000101cf20 in mbox_static_message (r=0x60000000004b9418,
f=0x60000000003d0000) at mod_mbox_out.c:1100
conf = (mbox_dir_cfg_t *) 0x60000000002e0a28
m = (Message *) 0x60000000003d03e8
baseURI = 0x60000000003d00b0 "/mod_mbox/db-derby-dev/200507.mbox"
from = 0xc00000000000038e <Address 0xc00000000000038e out of
bounds>
context = (char **) 0x60000000004bb217
msgID = 0x60000000004bb217 "<42...@Sourcery.Org>"
#3 0x2000000001018b50 in mbox_file_handler (r=0x60000000004b9418)
at mod_mbox_file.c:231
f = (apr_file_t *) 0x60000000003d0000
fi = {pool = 0x60000000004b93a8, valid = 7598448, protection =
1604,
filetype = APR_REG, user = 504, group = 5034, inode = 3408325,
device = 2065, nlink = 1, size = 13317914, csize = 4294967296,
atime = 1129164697000000, mtime = 1123654438000000,
ctime = 1128692829000000,
fname = 0x60000000003076c8
"/x1/mail-archives.apache.org/mod_mbox/db-derby-dev/200507.mbox", name =
0x0, filehand = 0x0}
status = 4953400
#4 0x400000000004fd70 in ap_run_handler (r=0x60000000004b9418) at
config.c:158
>
>> * It can be built against httpd release 2.{0,2}.y without
>> modifications to either httpd, mod_mbox source code or build files
>
> We got this point.
>
>> * There is documentation that allows a user (as opposed to the
>> person who wrote the code) to install it and get started serving
>> mail archives
>
> Documentation is currently inexistant, but if we choose to make a
> mod_mbox release, I could do it in the next few hours. I still have my
> .xml file from my last failed attempt on providing documentation (it
> was rejected because mod_mbox was not part of the main distribution).
>
>> * There are Z number of open bugs in Bugzilla against the module and
>> T of those need to be fixed before we can release, while U of them
>> can be waived
>
> Altough ASF's Bugzilla does not have a 'mod_mbox' project and no bugs
> are currently reported for mod_mbox to the Apache-2.x bugzilla
> project, the STATUS file is kinda verbose on known bugs and
> incompatibilities.
>
> But the fact is that we came to a running mod_mbox (server side) and
> browser incompatibilities are avoided by deactivating the dynamic
> browser if the client is not compatible.
>
>> S. (and why 0.2, why not 1.0? What are the criteria for 1.0?)
>
> That's why I only want to call it 0.2 and not 1.0. Because a 0.3 will
> come in the next months I hope to fix these problems and improve the
> thing. I don't think we should call it 1.0 until we make the dynamic
> browser work everywhere. Google makes it for every single
> bleeding-edge web-based application they do, why not us ?
>
> Thanks for the reply,
> - Sam
>
Re: A mod_mbox release ?
Posted by Maxime Petazzoni <ma...@bulix.org>.
Hi,
* Sander Temme <sc...@apache.org> [2005-12-11 20:06:54]:
> Maxime, folks,
>
> On Dec 11, 2005, at 4:45 PM, Maxime Petazzoni wrote:
>
> > I'm not yet used to release management (even for a small module like
> > mod_mbox), so I'll be very pleased to get some feedback, comments and
> > pointers if we decide to make a 0.2 release for mod_mbox !
>
> Good idea: releasing the code in packaged form should encourage its use.
>
> The main aspect of Release Management is (all IMHO of course) making
> an informed judgement of whether the current code base is "good
> enough" to release. This includes data points like:
>
> * It's been running on mail-archives.apache.org with fewer than X
> cores (with X tending to 0)
I don't know if I'm able to check this point on my own : where do
coredumps go ? Do I have enough access rights to check for them ?
Anyway, since my last fixes against core dumps, you did not report any
of them.
> * It can be built against httpd release 2.{0,2}.y without
> modifications to either httpd, mod_mbox source code or build files
We got this point.
> * There is documentation that allows a user (as opposed to the
> person who wrote the code) to install it and get started serving
> mail archives
Documentation is currently inexistant, but if we choose to make a
mod_mbox release, I could do it in the next few hours. I still have my
.xml file from my last failed attempt on providing documentation (it
was rejected because mod_mbox was not part of the main distribution).
> * There are Z number of open bugs in Bugzilla against the module and
> T of those need to be fixed before we can release, while U of them
> can be waived
Altough ASF's Bugzilla does not have a 'mod_mbox' project and no bugs
are currently reported for mod_mbox to the Apache-2.x bugzilla
project, the STATUS file is kinda verbose on known bugs and
incompatibilities.
But the fact is that we came to a running mod_mbox (server side) and
browser incompatibilities are avoided by deactivating the dynamic
browser if the client is not compatible.
> S. (and why 0.2, why not 1.0? What are the criteria for 1.0?)
That's why I only want to call it 0.2 and not 1.0. Because a 0.3 will
come in the next months I hope to fix these problems and improve the
thing. I don't think we should call it 1.0 until we make the dynamic
browser work everywhere. Google makes it for every single
bleeding-edge web-based application they do, why not us ?
Thanks for the reply,
- Sam
--
Maxime Petazzoni (http://www.bulix.org)
-- gone crazy, back soon. leave message.
Re: A mod_mbox release ?
Posted by Sander Temme <sc...@apache.org>.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Maxime, folks,
On Dec 11, 2005, at 4:45 PM, Maxime Petazzoni wrote:
> I'm not yet used to release management (even for a small module like
> mod_mbox), so I'll be very pleased to get some feedback, comments and
> pointers if we decide to make a 0.2 release for mod_mbox !
Good idea: releasing the code in packaged form should encourage its use.
The main aspect of Release Management is (all IMHO of course) making
an informed judgement of whether the current code base is "good
enough" to release. This includes data points like:
* It's been running on mail-archives.apache.org with fewer than X
cores (with X tending to 0)
* It can be built against httpd release 2.{0,2}.y without
modifications to either httpd, mod_mbox source code or build files
* There is documentation that allows a user (as opposed to the person
who wrote the code) to install it and get started serving mail archives
* There are Z number of open bugs in Bugzilla against the module and
T of those need to be fixed before we can release, while U of them
can be waived
Based on that information we can decide to release and make an
announcememt.
S. (and why 0.2, why not 1.0? What are the criteria for 1.0?)
- --
sander@temme.net http://www.temme.net/sander/
PGP FP: 51B4 8727 466A 0BC3 69F4 B7B8 B2BE BC40 1529 24AF
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)
iD8DBQFDnPdinjkrwtLH+RIRAjWhAJ9GKiiZL3ve+hwPeJu0tPQE3lhNPwCfV44j
5VH+MgycdHZ9Tw3tmJMQhvE=
=1rEW
-----END PGP SIGNATURE-----
Re: A mod_mbox release ?
Posted by Eli Marmor <ma...@netmask.it>.
Maxime Petazzoni wrote:
> ...
>
> I'm not yet used to release management (even for a small module like
> mod_mbox), so I'll be very pleased to get some feedback, comments and
> pointers if we decide to make a 0.2 release for mod_mbox !
I think Sam (=Maxime) is too modest to ask for the inclusion of mod_mbox
in the "modules/experimental" directory httpd; I think that this was the
original purpose of giving it to SoC, wasn't it?
And I'm sure that Sam will agree to promise to agree for future axing of
his module, IF it will not prove the benefit and quality that are
expected from a standard module of Apache.
In any case, thank you Sam!
--
Eli Marmor
marmor@netmask.it
Netmask (El-Mar) Internet Technologies Ltd.
__________________________________________________________
Tel.: +972-9-766-1020 8 Yad-Harutzim St.
Fax.: +972-9-766-1314 P.O.B. 7004
Mobile: +972-50-5237338 Kfar-Saba 44641, Israel