On 17/12/14 16:31 , "José Bollo" <[email protected]> wrote:

>Le mercredi 17 décembre 2014 à 13:55 +0000, Kanevskiy, Alexander a
>écrit :
>
>> - one of the most important things: no broken builds should be in
>> profiles. Failed to build package is one of the first priority
>>candidates
>> to be selected for build if any meta information in repository changes.
>> so, system can spend times on trying to build something that is broken
>>and
>> that is not unblocking packages further in the chain. In Common at the
>> moment there are 5 build failures (at least 2 package changes seems to
>>be
>> accepted without any testing).
>
>Hello,
>
>You are right. Please accept my apologies for that state but also accept
>some explanations.
>
>First, croswwalk was broken because of some infrastructure issue. As you
>know it is a huge piece of code and it reached the 3Gb as .tar. Ronan Le
>Martret found that replacing .tar by .tar.gz in the spec file solves the
>issue encountered this morning.

there was a patch in webkit-efl era that was done by the same people as
current crosswalk developers, but somehow now lost in newly done packaging:


in crosswalk source, in .gitattributes file it should be lines similar to
those:

-8<-
# For package build time/size optimization purposes, do not export Test
data.
/LayoutTests export-ignore
/ManualTests export-ignore
/PerformanceTests export-ignore

-8<-

those would skip packaging a huge pile of data that is not needed anywhere
during build or normal usage/development of crosswalk inside Tizen.
And that would bring source tarball few gigabytes less in size.


>
>Second, last week Stephane changed the default compiling flags. I guess
>that the new flags are in place since dec. 12th. This has revealed some
>new warnings that are now breaking packages that have the option -Werror
>set and that was okay before. I proposed 2 changes for solving this
>issues [1] and [2].

Generally, those changes must be done in separate trial projects before
pushed to production.
And changes to fix build failures also should be done before it’s pushed
to production.
It must be nothing more important than broken production.


>
>Third, during the transition between me and Stephane, Crosswalk was
>broken and some decision about accepting packages or not was hard to
>take. (And, consequently, the submit queue also grew).
>
>If I had maintenance right on prerealease, I would stop selectively
>their build until recovery of normal load.

manual intervention on workflow or bending/ignoring rules and
recommendations about warnings/errors is exactly reasons of abnormal
behaviour of the systems.

>
>Best regards
>José Bollo
>
>[1] https://review.tizen.org/gerrit/#/c/32354/
>[2] https://review.tizen.org/gerrit/#/c/32362/
>
>
>


-- 
Best regards, Alexander Kanevskiy.



---------------------------------------------------------------------
Intel Finland Oy
Registered Address: PL 281, 00181 Helsinki 
Business Identity Code: 0357606 - 4 
Domiciled in Helsinki 

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.
_______________________________________________
Dev mailing list
[email protected]
https://lists.tizen.org/listinfo/dev

Reply via email to