What is needed:
- apply the patch
- run rpm test suite
- rebuild perl-URPM with the two patches from cauldron
- run perl-URPM testsuite

Assignee: bugsquad => dmorganec

I've a mga1 branch ready to dcommit

See http://svnweb.mageia.org/soft/rpm/perl-URPM/branches/1/

URL: (none) => http://svnweb.mageia.org/soft/rpm/perl-URPM/branches/1/

Ah thanks, I was waiting for your commit which was ready :)
I will try to build a package and test it tonight.

I'll push 2 more fixes in order to have better filesizes in synthesis

rpm test suite is not very helpful.

Before patching: 6 12 29 30 31 32 33 34 35 36 37 45 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 failed
After patching:  6 12 29 30 31 32 33 34 35 36 37 45 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 failed

So, no change but still 35/80 tests failing :(

:-)

I just pushed perl-URPM-3.38.1.1-1.mga1
You should also compare urpmi's testsuite with rpm/perl-URPM from core/release+updates and then with updates_testing

Note that urpmi testsuite don't pass since 2010.0 or 2010.1.
I've fixed it in cauldron.
I can backport testsuite fixes though...

I have verified that the genhdlist2 problem is fixed with perl-URPM-3.38.1.1-1.mga1.

I will try to do more complete testing tomorrow.

I just submited urpmi with backported testsuite fixes

I also submited rpmtools with a fixed genhdlist2 (needing the new URPM) so that package sizes be correct in synthesis (shame on pixel...)

Since servers building cauldron/mga2's synthesis will be using mga1, this will fix cauldron's synthesis regarding this

What package contaings the rpm test suite?

CC: (none) => davidwhodgins

The SRPM of rpm

Tested with testsuite + quite heavy use on my x86_64 machine:

perl-URPM-3.38.1.1-2.mga1
urpmi-6.40.1-1.mga1
gurpmi-6.40.1-1.mga1

rpmtools-6.2-1.mga1
genhdlist2-6.2-1.mga1

rpm-build-4.8.1-10.3.mga1
rpm-4.8.1-10.3.mga1
lib64rpm-devel-4.8.1-10.3.mga1
lib64rpm1-4.8.1-10.3.mga1

Not tested: python-rpm-4.8.1-10.3.mga1

I have also confirmed that genhdlist2 can now reuse all existing hdlist.

I don't know what was the package size bug and how to reproduce/test it.

Is this bug ready for the QA ?
https://wiki.mageia.org/en/Updates_policy#Maintainer_.28or_any_interested_packager.29

I think it is but we should describe how to reproduce and test the known fixes.

(In reply to comment #12)
> I don't know what was the package size bug and how to reproduce/test it.

zcat media_info/sythesis.cz>/tmp/a1
check some packages file sizes, see they don't match real package sizes.
remove hdlists & synthesis, regenerate them, see that package file sizes are now OK.

Created attachment 1705 [details]
Output showing testing of genhdlist

I'm not seeing any difference in genhdlist2 processing.

There is something wrong with your test, did you install all the packages?

"adding 177 new rpms not available in existing hdlist" should not happen.

Thanks, I had missed some.  Used --auto-select with updates testing
enabled to get them. Now I get ...
# time genhdlist2 .
filtering ./media_info/hdlist.cz into hdlist.cz.tmp
adding 6 new rpms not available in existing hdlist
replacing ./media_info/hdlist.cz with hdlist.cz.tmp
replacing ./media_info/synthesis.hdlist.cz with synthesis.hdlist.cz.tmp
updating ./media_info/MD5SUM

real    0m6.135s
user    0m1.630s
sys     0m0.200s

For the file size, one example ...
-rw-r--r-- 1 root root   866195 Dec 29 23:54 dhcp-server-4.2.1-0.P1.3.1.mga1.i586.rpm

In the before it has
@filesize@865251
In the after it has
@filesize@866195

Testing complete on i586 for the srpms
perl-URPM-3.38.1.1-2.mga1.src.rpm
urpmi-6.40.1-1.mga1.src.rpm
rpmtools-6.2-1.mga1.src.rpm
rpm-4.8.1-10.3.mga1.src.rpm

Am I missing any srpms?

No that's it, I will check the filesize fix on x86_64.

-rw-r--r-- 1 pterjan pterjan 4893 mars  20  2011 /mnt/mirror/Mageia/distrib/cauldron/x86_64/media/nonfree/release/rt73-firmware-1.8-6.mga1.noarch.rpm

Before:
@filesize@3949
@info@rt73-firmware-1.8-6.mga1.noarch@0@2074@System/Kernel and hardware

After:
@filesize@4893
@info@rt73-firmware-1.8-6.mga1.noarch@0@2074@System/Kernel and hardware

I confirm this bug is also fixed.

Please take a look at bug 4855, which is for urpmi in cauldron.

Is there much of a difference between cauldron and this version?

Anyone object to validating this update or would you rather
have more testing.  These are critical packages.

Hopefully once it'll lands in mga1, we'll get proper sizes estimation in urpmi in rpmdrake

I think it is ready

Validating the update.

Could someone from the sysadmin team push the srpms
perl-URPM-3.38.1.1-2.mga1.src.rpm
urpmi-6.40.1-1.mga1.src.rpm
rpmtools-6.2-1.mga1.src.rpm
rpm-4.8.1-10.3.mga1.src.rpm
from Core Updates Testing to Core Updates.

Advisory: The bugfix update corrects errors in the generation of
hdlists that resulted in incorrect rpm package sizes being stored
in the hdlist files, and caused hdlist files to be regenerated
from scratch, when adding new rpm packages.

https://bugs.mageia.org/show_bug.cgi?id=4613

Keywords: (none) => validated_update
CC: (none) => sysadmin-bugs

update pushed

Status: ASSIGNED => RESOLVED
CC: (none) => tmb
Resolution: (none) => FIXED

Why was this never assigned to QA?

(In reply to comment #27)
> Why was this never assigned to QA?

Sorry.  Thought it was in comment 14.  As I'd already posted
a comment, I was already on the cc list.  My mistake for
validating the update without checking that it was assigned
to qa first.

It isn't really our responsibility to assign them to us Dave :)

https://wiki.mageia.org/en/Updates_policy#Maintainer_.28or_any_interested_packager.29

(In reply to comment #29)
> It isn't really our responsibility to assign them to us Dave :)
> 
> https://wiki.mageia.org/en/Updates_policy#Maintainer_.28or_any_interested_packager.29

Agreed, however it was my mistake to go ahead with testing, and then
validating the update, without checking to see if it was assigned
to qa.