Bug 130236 - libpbl-1.03 (databases/libpbl) is broken, spamprobe-1.4d_1 (mail/spamprobe) does not work at all with a pbl file
Summary: libpbl-1.03 (databases/libpbl) is broken, spamprobe-1.4d_1 (mail/spamprobe) d...
Status: Closed FIXED
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: Any Any
: Normal Affects Only Me
Assignee: Stefan Walter
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2009-01-06 15:40 UTC by kenji.rikitake
Modified: 2009-10-12 12:15 UTC (History)
0 users

See Also:


Attachments
libpbl.update.diff (2.17 KB, patch)
2009-10-01 17:19 UTC, consulting
no flags Details | Diff
patch-pbl (1.01 KB, text/plain)
2009-10-01 17:19 UTC, consulting
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description kenji.rikitake 2009-01-06 15:40:01 UTC
port libpbl-1.03 (databases/libpbl) no longer works properly on 7.1-RELEASE.

persistent warnings found when building libpbl-1.03:
===>  Vulnerability check disabled, database not found
===>  Extracting for libpbl-1.03
=> MD5 Checksum OK for pbl_1_03.tar.gz.
=> SHA256 Checksum OK for pbl_1_03.tar.gz.
===>  Patching for libpbl-1.03
===>  Configuring for libpbl-1.03
===>  Building for libpbl-1.03
cc -O2 -fno-strict-aliasing -pipe  -Wall -I. -c pbl.c
cc -O2 -fno-strict-aliasing -pipe  -Wall -I. -c pblhash.c
cc -O2 -fno-strict-aliasing -pipe  -Wall -I. -c pblkf.c
pblkf.c: In function 'pblKfCreate':
pblkf.c:3820: warning: pointer targets in assignment differ in signedness
pblkf.c:3824: warning: pointer targets in assignment differ in signedness
pblkf.c:3832: warning: pointer targets in passing argument 2 of 'pblDataAppend' differ in signedness
pblkf.c: In function 'pblSplit':
pblkf.c:4376: warning: pointer targets in initialization differ in signedness
pblkf.c:4502: warning: pointer targets in assignment differ in signedness
pblkf.c: In function 'pblKfInsert':
pblkf.c:4864: warning: pointer targets in passing argument 2 of 'pblDataAppend' differ in signedness
pblkf.c: In function 'pblKfUpdate':
pblkf.c:5467: warning: pointer targets in passing argument 1 of 'pblParamsCheck' differ in signedness
pblkf.c:5526: warning: pointer targets in passing argument 2 of 'pblDataWrite' differ in signedness
pblkf.c: In function 'pblKfFind':
pblkf.c:5892: warning: pointer targets in passing argument 3 of 'pblParamsCheck' differ in signedness
pblkf.c: In function 'pblKfRead':
pblkf.c:6013: warning: pointer targets in passing argument 1 of 'pblParamsCheck' differ in signedness
pblkf.c:6013: warning: pointer targets in passing argument 3 of 'pblParamsCheck' differ in signedness
pblkf.c:6064: warning: pointer targets in passing argument 2 of 'pblDataGet' differ in signedness
pblkf.c: In function 'pblKfBlockPrint':
pblkf.c:6503: warning: pointer targets in assignment differ in signedness
cc -O2 -fno-strict-aliasing -pipe  -Wall -I. -c pblisam.c
pblisam.c: In function 'pblRKey2MainKey':
pblisam.c:174: warning: pointer targets in passing argument 2 of 'pbl_VarBufToLong' differ in signedness
pblisam.c:188: warning: pointer targets in passing argument 1 of 'snprintf' differ in signedness
pblisam.c:218: warning: pointer targets in passing argument 2 of 'pbl_VarBufToLong' differ in signedness
pblisam.c:240: warning: pointer targets in passing argument 1 of 'snprintf' differ in signedness
pblisam.c: In function 'pblMainKey2RKey':
pblisam.c:310: warning: pointer targets in passing argument 1 of 'strtoul' differ in signedness
pblisam.c:312: warning: pointer targets in passing argument 1 of 'strtoul' differ in signedness
pblisam.c:317: warning: pointer targets in passing argument 1 of 'strtoul' differ in signedness
pblisam.c: In function 'pblIsamDupKeyLen':
pblisam.c:341: warning: pointer targets in passing argument 1 of 'pblRKey2MainKey' differ in signedness
pblisam.c: In function 'pblIsamDupKeyCompare':
pblisam.c:385: warning: pointer targets in passing argument 3 of 'pblRKey2MainKey' differ in signedness
pblisam.c:386: warning: pointer targets in passing argument 3 of 'pblRKey2MainKey' differ in signedness
pblisam.c: In function 'pblIsamSetCurrentRecord':
pblisam.c:1040: warning: pointer targets in passing argument 3 of 'pblKfFind' differ in signedness
pblisam.c: In function 'pblIsamInsert':
pblisam.c:1194: warning: pointer targets in passing argument 1 of 'pblMainKey2RKey' differ in signedness
pblisam.c:1194: warning: pointer targets in passing argument 3 of 'pblMainKey2RKey' differ in signedness
pblisam.c:1331: warning: pointer targets in passing argument 4 of 'pblKfInsert' differ in signedness
pblisam.c:1373: warning: pointer targets in passing argument 2 of 'pblKfInsert' differ in signedness
pblisam.c:1398: warning: pointer targets in passing argument 2 of 'pblKfInsert' differ in signedness
pblisam.c: In function 'pblIsamDelete':
pblisam.c:1539: warning: pointer targets in passing argument 1 of 'pblMainKey2RKey' differ in signedness
pblisam.c:1539: warning: pointer targets in passing argument 3 of 'pblMainKey2RKey' differ in signedness
pblisam.c:1624: warning: pointer targets in passing argument 3 of 'pblKfFind' differ in signedness
pblisam.c:1636: warning: pointer targets in passing argument 3 of 'pblKfFind' differ in signedness
pblisam.c: In function 'pblIsamGetMainKey':
pblisam.c:1720: warning: pointer targets in passing argument 2 of 'pblKfRead' differ in signedness
pblisam.c:1730: warning: pointer targets in passing argument 1 of 'pblRKey2MainKey' differ in signedness
pblisam.c:1739: warning: pointer targets in passing argument 1 of 'strlen' differ in signedness
pblisam.c: In function 'pblIsamFindDupKey':
pblisam.c:1798: warning: pointer targets in passing argument 3 of 'pblLongs2RKey' differ in signedness
pblisam.c:1826: warning: pointer targets in passing argument 3 of 'pblLongs2RKey' differ in signedness
pblisam.c:1838: warning: pointer targets in passing argument 3 of 'pblLongs2RKey' differ in signedness
pblisam.c:1865: warning: pointer targets in passing argument 3 of 'pblLongs2RKey' differ in signedness
pblisam.c:1898: warning: pointer targets in passing argument 3 of 'pblKfFind' differ in signedness
pblisam.c:1898: warning: pointer targets in passing argument 5 of 'pblKfFind' differ in signedness
pblisam.c: In function 'pblIsamFind':
pblisam.c:2052: warning: pointer targets in passing argument 5 of 'pblIsamGetMainKey' differ in signedness
pblisam.c: In function 'pblIsamThisKey':
pblisam.c:2172: warning: pointer targets in passing argument 1 of 'pblMainKey2RKey' differ in signedness
pblisam.c:2172: warning: pointer targets in passing argument 3 of 'pblMainKey2RKey' differ in signedness
pblisam.c: In function 'pblIsamGet':
pblisam.c:2310: warning: pointer targets in passing argument 3 of 'pblKfGetRel' differ in signedness
pblisam.c:2359: warning: pointer targets in passing argument 3 of 'pblKfGetRel' differ in signedness
pblisam.c:2377: warning: pointer targets in passing argument 3 of 'pblKfGetAbs' differ in signedness
pblisam.c:2424: warning: pointer targets in passing argument 3 of 'pblKfGetAbs' differ in signedness
pblisam.c:2476: warning: pointer targets in passing argument 5 of 'pblIsamGetMainKey' differ in signedness
pblisam.c:2491: warning: pointer targets in passing argument 1 of 'pblIsamDupKeyLen' differ in signedness
pblisam.c: In function 'pblIsamUpdateData':
pblisam.c:2820: warning: pointer targets in passing argument 3 of 'pblKfFind' differ in signedness
pblisam.c:2856: warning: pointer targets in passing argument 3 of 'pblKfFind' differ in signedness
pblisam.c:2931: warning: pointer targets in passing argument 2 of 'pblKfInsert' differ in signedness
pblisam.c:2945: warning: pointer targets in passing argument 3 of 'pblKfFind' differ in signedness
pblisam.c: In function 'pblIsamUpdateKey':
pblisam.c:3020: warning: pointer targets in passing argument 4 of 'pblIsamGet' differ in signedness
pblisam.c:3175: warning: pointer targets in passing argument 1 of 'pblMainKey2RKey' differ in signedness
pblisam.c:3175: warning: pointer targets in passing argument 3 of 'pblMainKey2RKey' differ in signedness
pblisam.c:3269: warning: pointer targets in passing argument 1 of 'pblMainKey2RKey' differ in signedness
pblisam.c:3269: warning: pointer targets in passing argument 3 of 'pblMainKey2RKey' differ in signedness
pblisam.c:3297: warning: pointer targets in passing argument 4 of 'pblKfInsert' differ in signedness
/usr/bin/ar rc libpbl.a pbl.o pblhash.o pblkf.o pblisam.o
/usr/bin/ar ts libpbl.a

Fix: 

Stopping using the libpbl for the spamprobe will solve the problem.

I suggest the maintainer of spamprobe to add
WITHOUT_PBL=true
as a build-in option in the Makefile to prevent further bugs showing up.

And I also suggest the port maintainter to declare
libpbl-1.03 is BROKEN.
How-To-Repeat: spamprobe-1.4d_1 (port mail/spamprobe) no longer works on a PBL file.
See PR ports/119180 for the details.
Comment 1 Edwin Groothuis freebsd_committer freebsd_triage 2009-01-06 15:40:09 UTC
Maintainer of databases/libpbl,

Please note that PR ports/130236 has just been submitted.

If it contains a patch for an upgrade, an enhancement or a bug fix
you agree on, reply to this email stating that you approve the patch
and a committer will take care of it.

The full text of the PR can be found at:
    http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/130236

-- 
Edwin Groothuis via the GNATS Auto Assign Tool
edwin@FreeBSD.org
Comment 2 Edwin Groothuis freebsd_committer freebsd_triage 2009-01-06 15:40:11 UTC
State Changed
From-To: open->feedback

Awaiting maintainers feedback (via the GNATS Auto Assign Tool)
Comment 3 Stefan Walter freebsd_committer freebsd_triage 2009-01-08 10:44:38 UTC
Responsible Changed
From-To: freebsd-ports-bugs->stefan

Take.
Comment 4 Stefan Walter freebsd_committer freebsd_triage 2009-01-09 08:23:19 UTC
Hi Kenji,

I maintain mail/spamprobe. On my 7.1-RELEASE i386 system, I get the same
warnings when compiling libpbl. However, I use spamprobe with a libpbl
database all the time and still cannot reproduce any problem with it. (I
just rebuilt libpbl and spamprobe and created a new database to be sure -
no problem here.)

Can you please explain in detail what you are doing and what problems you
are facing so I can try to reproduce that?

Regards,
Stefan
Comment 5 kenji.rikitake 2009-01-09 08:46:29 UTC
Hello Stefan:

I was trying to open a very large pbl database (approx. 200Mbytes) and I
could no longer open it anymore, after I upgraded the system from
6.4-RELEASE to 7.1-RELEASE on i386.  I'll try to retrieve the database I
was using from my past backup.  Give me a little time.

Regards,
Kenji Rikitake
Comment 6 kenji.rikitake 2009-01-09 09:17:11 UTC
Hello Stefan:

I got the database back from the past archive.
The error is reproduced as follows.

reseaux[1]% pwd
/var/tmp/kenji/.spamprobe
reseaux[2]% ls -l
total 216656
-rw-------  1 kenji  k2r          0 Jan  9 18:01 lock
-rw-------  1 kenji  k2r  221712384 Dec 31 18:55 sp_words.pkey
reseaux[3]% spamprobe-withpbl -d /var/tmp/kenji/.spamprobe/ export
caught runtime exception: pblKfGetAbs: Cannot allocate memory (12)
reseaux[4]% 

The tcsh limit command says:

reseaux[13]% limit
cputime      unlimited
filesize     unlimited
datasize     524288 kbytes
stacksize    65536 kbytes
coredumpsize 0 kbytes
memoryuse    unlimited
vmemoryuse   unlimited
descriptors  14781 
memorylocked unlimited
maxproc      7390 
sbsize       unlimited

The error was same as I compiled the libpbl-1.03 and spamprobe-1.4d_1
by 7.1-RELEASE-stock gcc-4.2.1 and gcc42 (gcc 4.2.5 20081126).

I will try what happens when the two ports are compiled with gcc34,
but I need to build the gcc34 by myself first, so please give me a time
again.

Regards,
Kenji Rikitake

PS: I was a bit shocked when I found out the 220Mbyte database content
was no longer useful.  That was the reason I classified the PR "critical."
Comment 7 Stefan Walter freebsd_committer freebsd_triage 2009-01-09 09:19:38 UTC
Kenji Rikitake, 09.01.09, 09:46h CET:

> Hello Stefan:
> 
> I was trying to open a very large pbl database (approx. 200Mbytes) and I
> could no longer open it anymore, after I upgraded the system from
> 6.4-RELEASE to 7.1-RELEASE on i386.  I'll try to retrieve the database I
> was using from my past backup.  Give me a little time.

Sure, thanks for your feedback! Since you mentioned a large pbl db, it
might be you're hitting a limit there - see [1]. If you find out anything,
let me know.

Regards,
Stefan

[1]: http://sourceforge.net/mailarchive/forum.php?thread_name=495E45E8.5060800%40schuerger.com&forum_name=spamprobe-users
Comment 8 kenji.rikitake 2009-01-09 09:30:57 UTC
Stefan: thanks for the info.  I'll try rebuilding the libpbl-1.03
with the -DPBLFILEBLOCKS=0x3ffff option.  Retrieving the past spam
database is worth trying. Regards, Kenji Rikitake
Comment 9 kenji.rikitake 2009-01-09 10:07:39 UTC
Hello again Stefan:

-DPBLFILEBLOCKS=0x3ffff for libpbl didn't work with stock gcc 4.2.1.

I built the gcc34 from the port.

I tried building the libpbl-1.03 and spamprobe-1.4d with gcc34
by specifying USE_GCC=3.4 in the port Makefiles.
The result was as follows
(to extract the 220Mbyte database with "spamprobe export"):

(libpbl in stock gcc 4.2.1) and (spamprobe in stock gcc 4.2.1): didn't work
(libpbl in gcc34) and (spamprobe in stock gcc 4.2.1): didn't work
(libpbl in gcc34) and (spamprobe in gcc34): worked!

I suspect there's something different on the memory allocation
between gcc34 and stock gcc 4.2.1.

All the above cases are tested on 7.1-RELEASE i386.

FYI.
Kenji Rikitake
Comment 10 Stefan Walter freebsd_committer freebsd_triage 2009-01-12 07:59:34 UTC
Kenji Rikitake, 09.01.09, 11:07h CET:

> Hello again Stefan:
> 
> -DPBLFILEBLOCKS=0x3ffff for libpbl didn't work with stock gcc 4.2.1.
> 
> I built the gcc34 from the port.
> 
> I tried building the libpbl-1.03 and spamprobe-1.4d with gcc34
> by specifying USE_GCC=3.4 in the port Makefiles.
> The result was as follows
> (to extract the 220Mbyte database with "spamprobe export"):
> 
> (libpbl in stock gcc 4.2.1) and (spamprobe in stock gcc 4.2.1): didn't work
> (libpbl in gcc34) and (spamprobe in stock gcc 4.2.1): didn't work
> (libpbl in gcc34) and (spamprobe in gcc34): worked!
> 
> I suspect there's something different on the memory allocation
> between gcc34 and stock gcc 4.2.1.
> 
> All the above cases are tested on 7.1-RELEASE i386.

Thanks for the info! I'll look at this a bit closer soon (hopefully...).

Regards,
Stefan
Comment 11 Stefan Walter freebsd_committer freebsd_triage 2009-01-27 07:30:18 UTC
Hi Kenji, hi Zach,

I suspect the real problem are the "differ in signedness" errors that you
see when compiling libpbl with gcc 4.2.1. I guess that the declaration of
some variables in the libpbl source code as 'signed' or 'unsigned'
(explicitly or implicitly) is incorrect, so the solution would be to fix
those declarations.

At the moment, it doesn't look like I'll have the time to work on this
soon, but if anyone of you is willing to work on this, I'd be willing to
check patches and commit them.

Best regards,
Stefan
Comment 12 hideo 2009-01-28 03:55:42 UTC
Stefan Walter (Tue 01/27/09 08:30):
> Hi Kenji, hi Zach,
> 
> I suspect the real problem are the "differ in signedness" errors that you
> see when compiling libpbl with gcc 4.2.1. I guess that the declaration of
> some variables in the libpbl source code as 'signed' or 'unsigned'
> (explicitly or implicitly) is incorrect, so the solution would be to fix
> those declarations.
> 
> At the moment, it doesn't look like I'll have the time to work on this
> soon, but if anyone of you is willing to work on this, I'd be willing to
> check patches and commit them.
> 
> Best regards,
> Stefan

Though I can "fix" signedness warnings I don't have time to verify
that the hundreds of casts necessary won't break something else. I can
see if the author is interested in updating it for gcc 4+ but for the
time being perhaps it's best if we mark it broken.  Shall I submit a
patch for this?

Zach
Comment 13 Stefan Walter freebsd_committer freebsd_triage 2009-01-28 07:35:28 UTC
Hi Zach,

Zach Thompson, 28.01.09, 04:55h CET:

> Though I can "fix" signedness warnings I don't have time to verify
> that the hundreds of casts necessary won't break something else. I can
> see if the author is interested in updating it for gcc 4+ but for the
> time being perhaps it's best if we mark it broken.  Shall I submit a
> patch for this?

hm...it's only broken with gcc 4.x as it seems, so adding USE_GCC=3.4
would help libpbl - I don't see any signedness warnings there, and Kenji
reported spamprobe works in that case (when compiled with gcc 3.4, too).
On the other hand, I'm a bit reluctant to add the necessary dependency on
gcc 3.4 to the spamprobe port since it's pretty heavy-weight. Furthermore,
the problem seems to exist with databases >200 megabytes only; mine
doesn't grow beyond ~35 megabytes.

My suggestion would be the following:

- I mark databases/libpbl as broken for now.
- You contact libpbl's author and see if we can get a fix for the problem
  soon. (There aren't that many signedness warnings, so maybe we're lucky
  and it's not too much work.)
- If we get a fix, we apply it; if not, we'll add a dependency on gcc 3.4
  to libpbl and spamprobe (when compiled with support for libpbl).

Can we agree on this for now?

Best regards,
Stefan
Comment 14 kenji.rikitake 2009-01-28 08:59:13 UTC
Stefan and all:

I think this decision is rational and acceptable.
One clarification:
spamprobe without libppl works OK with gcc 4.x.

Regards,
Kenji Rikitake

In the message <20090128073528.GA2300@kyuzo.dunkelkammer.void>
dated Wed, Jan 28, 2009 at 08:35:04AM +0100,
Stefan Walter <stefan@FreeBSD.org> writes:
> My suggestion would be the following:
> 
> - I mark databases/libpbl as broken for now.
> - You contact libpbl's author and see if we can get a fix for the problem
>   soon. (There aren't that many signedness warnings, so maybe we're lucky
>   and it's not too much work.)
> - If we get a fix, we apply it; if not, we'll add a dependency on gcc 3.4
>   to libpbl and spamprobe (when compiled with support for libpbl).
> 
> Can we agree on this for now?
> 
> Best regards,
> Stefan
Comment 15 hideo 2009-01-28 13:48:57 UTC
Stefan Walter (Wed 01/28/09 08:35):
> Hi Zach,
> 
> Zach Thompson, 28.01.09, 04:55h CET:
> 
> > Though I can "fix" signedness warnings I don't have time to verify
> > that the hundreds of casts necessary won't break something else. I can
> > see if the author is interested in updating it for gcc 4+ but for the
> > time being perhaps it's best if we mark it broken.  Shall I submit a
> > patch for this?
> 
> hm...it's only broken with gcc 4.x as it seems, so adding USE_GCC=3.4
> would help libpbl - I don't see any signedness warnings there, and Kenji
> reported spamprobe works in that case (when compiled with gcc 3.4, too).
> On the other hand, I'm a bit reluctant to add the necessary dependency on
> gcc 3.4 to the spamprobe port since it's pretty heavy-weight. Furthermore,
> the problem seems to exist with databases >200 megabytes only; mine
> doesn't grow beyond ~35 megabytes.
> 
> My suggestion would be the following:
> 
> - I mark databases/libpbl as broken for now.
> - You contact libpbl's author and see if we can get a fix for the problem
>   soon. (There aren't that many signedness warnings, so maybe we're lucky
>   and it's not too much work.)
> - If we get a fix, we apply it; if not, we'll add a dependency on gcc 3.4
>   to libpbl and spamprobe (when compiled with support for libpbl).
> 
> Can we agree on this for now?
> 
> Best regards,
> Stefan

Stefan,

Sounds good to me.

Zach
Comment 16 dfilter service freebsd_committer freebsd_triage 2009-02-02 14:23:27 UTC
stefan      2009-02-02 14:23:18 UTC

  FreeBSD ports repository

  Modified files:
    databases/libpbl     Makefile 
  Log:
  Mark as BROKEN on OSVERSION >= 700000 for now: does not work with databases
  bigger than ~200 MB.
  
  PR:             130236
  Submitted by:   Kenji Rikitake <kenji.rikitake@acm.org>
  Approved by:    maintainer
  
  Revision  Changes    Path
  1.8       +7 -1      ports/databases/libpbl/Makefile
_______________________________________________
cvs-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/cvs-all
To unsubscribe, send any mail to "cvs-all-unsubscribe@freebsd.org"
Comment 17 Stefan Walter freebsd_committer freebsd_triage 2009-02-05 08:07:22 UTC
Hi Zach,

any news on this issue? Have you heard anything from libpbl's author, yet?

Regards,
Stefan
Comment 18 hideo 2009-02-05 13:32:49 UTC
Stefan Walter (Thu 02/05/09 09:07):
> Hi Zach,
> 
> any news on this issue? Have you heard anything from libpbl's author, yet?
> 
> Regards,
> Stefan

Hi Stefan,

Unfortunately I haven't heard from Peter yet.  Didn't have any luck
fixing the problem myself yet either, though I have gotten it to
compile without warnings.

Zach
Comment 19 hideo 2009-03-02 22:47:09 UTC
Zach Thompson (Thu 02/05/09 06:32):
> Stefan Walter (Thu 02/05/09 09:07):
> > Hi Zach,
> > 
> > any news on this issue? Have you heard anything from libpbl's author, yet?
> > 
> > Regards,
> > Stefan
> 
> Hi Stefan,
> 
> Unfortunately I haven't heard from Peter yet.  Didn't have any luck
> fixing the problem myself yet either, though I have gotten it to
> compile without warnings.
> 
> Zach

Just thought I'd pass along that I heard from the author (Peter Graf) 
of libpbl today.  He will look at getting it to work with gcc 4.2 but
probably won't be able to get back to me until the end of March.
 
Zach
Comment 20 Stefan Walter freebsd_committer freebsd_triage 2009-08-12 18:58:00 UTC
Hi Zach,

Zach Thompson, 02.03.09, 23:47h CET:

> Just thought I'd pass along that I heard from the author (Peter Graf) 
> of libpbl today.  He will look at getting it to work with gcc 4.2 but
> probably won't be able to get back to me until the end of March.

have you heard anything, yet?

Regards,
Stefan
Comment 21 Stefan Walter freebsd_committer freebsd_triage 2009-09-21 14:34:19 UTC
Hi Zach,

sorry for the very late reply. I somehow managed to defer this PR
repeatedly...

Zach Thompson, 12.08.09, 21:52h CEST:

> Yes, Peter updated libpbl to 1.04 and sent me an email back on March
> 15th.  I updated the port locally and tested.  However, I still got
> the same signedness errors and ultimately the same error when used via
> Spamprobe.  Though I sent him a follow-up I still haven't heard anything.

I guess this hasn't changed in the meantime? Was he able to reproduce the
problem?

> I guess I've done about as much as I can do on this port.  Perhaps
> someone with more C experience than me can figure out how to patch it.

Unfortunately, I haven't had much time for FreeBSD related work, and
probably won't be having any soon, so I still can't help here. Maybe
someone on the ports mailing list can help.

The best choice for those who have large databases would be to use one of
the other backends in spamprobe for the moment, I guess.

Regards,
Stefan
Comment 22 Stefan Walter freebsd_committer freebsd_triage 2009-09-25 18:08:36 UTC
Hello Peter,

at [1], there's a problem report concerning the FreeBSD port of PBL. It
looks like the problem described there is related to signedness issues
that somehow were not a problem with older gcc versions, but do result in
problems with more recent ones. I don't think it's FreeBSD-specific - it
should be reproducable on Linux systems, for instance, as well.

Could you have a look at the problem description in the report and maybe
provide some feedback regarding its solution?

Best regards,
Stefan

[1]: http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/130236
Comment 23 pgraf 2009-09-27 22:28:14 UTC
Hello Stefan

I will look into the matter, although last time I did, I could compile  
on all
current gcc compilers on Ubuntu, Suse and Mac OS without any warnings,
and somebody from your group still reported problems on FreeBSD.

Maybe this time I install FreeBSD myself.
but I know from my schedule for the next couple of weeks
that I will probably not have time in October, the week
starting November 9th is the most likely the first with
any spare time.

Hope that helps nevertheless

Greetings

Peter

Am 25.09.2009 um 19:08 schrieb Stefan Walter:

> Hello Peter,
>
> at [1], there's a problem report concerning the FreeBSD port of PBL.  
> It
> looks like the problem described there is related to signedness issues
> that somehow were not a problem with older gcc versions, but do  
> result in
> problems with more recent ones. I don't think it's FreeBSD-specific  
> - it
> should be reproducable on Linux systems, for instance, as well.
>
> Could you have a look at the problem description in the report and  
> maybe
> provide some feedback regarding its solution?
>
> Best regards,
> Stefan
>
> [1]: http://www.freebsd.org/cgi/query-pr.cgi?pr=ports/130236

----------------------------------------------------------
Peter Graf, Peter.Graf@mission-base.com
Tel: +49 (89) 2023 2149
Cell: +49 (170) 832 9632
https://www.xing.com/hp/Peter_Graf2/
http://www.mission-base.com/
skype: peter_graf


Comment 24 consulting 2009-10-01 17:19:46 UTC
    http://www.freebsd.org/cgi/query-pr.cgi?pr=130236

Please, try the attached update for the port. It:

   1. Updates the software to version 1.04 (the latest from author at
      the time of this writing).
   2. Ensures warning-free build.
   3. Declares the software as safe for parallel building.
   4. Runs the author's bundled self-tests after building.

I think, once the current urgent problem is resolved, the port should be
modified to build both static and shared version of the pbl-library
(using bsd.lib.mk), but that can wait.

I also dislike, how the databases/libpbl port edits the author's code on
the fly instead of static patches, which could've been sent to Peter
(CC-ed) to spare him some work -- such as port to MacOS, for which he
had to ifdef-out malloc.h, incorrectly assuming, that the include is
needed somewhere else... Changing this can also wait, though.

To build spamprobe against this new version of libpbl on my
amd64-system, I had to add the second attachment to mail/spamprobe/files
(it may not be needed on a 32-bit platform). My system is now busy
importing my giant (never purged) database into PBL-format from the BDB.
Things seem to work, but I'm yet to test thoroughly. Yours,

    -mi
Comment 25 Stefan Walter freebsd_committer freebsd_triage 2009-10-06 13:17:18 UTC
Hi all,

I tried Mikhail's patch for the the PBL and Spamprobe ports. I can confirm
that the compile time warnings with PBL are gone, and Spamprobe seems to
work with a freshly created and trained PBL database, too.

Zach: If you agree, I would commit it.

Best regards,
Stefan
Comment 26 dfilter service freebsd_committer freebsd_triage 2009-10-06 15:52:32 UTC
stefan      2009-10-06 14:52:01 UTC

  FreeBSD ports repository

  Modified files:
    databases/libpbl     Makefile distinfo 
    mail/spamprobe       Makefile 
  Added files:
    mail/spamprobe/files patch-src_database_FrequencyCBImpl_pbl.cc 
  Log:
  - Update databases/libpbl to 1.04, fix compile-time warnings, declare as safe
    for parallel building and run bundled self-tests after building.
  - Patch mail/spamprobe to make it work with libpbl on amd64.
  
  PR:             130236
  Submitted by:   Kenji Rikitake <kenji.rikitake@acm.org>
  Patch by:       mi
  Approved by:    Zach Thompson <hideo@lastamericanempire.com> (maintainer)
  
  Revision  Changes    Path
  1.9       +10 -13    ports/databases/libpbl/Makefile
  1.2       +3 -3      ports/databases/libpbl/distinfo
  1.26      +1 -1      ports/mail/spamprobe/Makefile
  1.1       +21 -0     ports/mail/spamprobe/files/patch-src_database_FrequencyCBImpl_pbl.cc (new)
_______________________________________________
cvs-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/cvs-all
To unsubscribe, send any mail to "cvs-all-unsubscribe@freebsd.org"
Comment 27 Stefan Walter freebsd_committer freebsd_triage 2009-10-12 12:14:47 UTC
State Changed
From-To: feedback->closed

Fixed in most recent update of libpbl/spamprobe.