[Bf-committers] call for test build 2

Tom M letterrip at gmail.com
Mon Mar 31 03:35:08 CEST 2008


Hi Kent,


>  Having said all that.  Are we doing a release or no?
>
>  If yes, I think it should be just like any other release, who cares if
>  its a bit under construction or not.  If we note it in the download page
>  its a non issue.  I think things are way to wishy washy right now and
>  everyone has no clue whats up with things.
>

at the start of Peach, the plan was to wait for 2.50, however, about
three or so months ago, it was realized that 1) 2.50 would take more
time to accomplish than was expected (this was determined in part as a
result from Ton's 'coding vacation') and that the soonest 2.50 could
be out would be after Siggraph; 2) That without an official release
any Blender user who wanted to make use of commercial render farms
could not do so (commercial render farms only allow official
releases); 3) That the lack of release was holding back books being
published 4) There would be a 'release' anyway since the version of
Blender used for Peach would ship on the CD.

Given those factors it was decided to go ahead and have a release
based on the code base used for Peach.

The longer testing cycle is due to having had such a long period of
time since the last 'real' release (2.45 was bug fix only - so it will
be almost a year since 2.44), and the substantial changes made in the
course of peach.


>  I think in general blender development is going great, I think the
>  last couple of releases though have dragged out a lot longer than all
>  of the others and it has made things harder not easier.  (I do think
>  there will be releases that take longer and special exceptions, however
>  this seems to be a trend in the wrong direction.  It could just
>  be me)
>

I agree that longer release cycles make things more difficult, with
major changes in a bunch of subsystems makes stabalization much more
difficult and the odds of bugs having interactions more likely.  I
agree that a trend of longer releases is the wrong direction, and I'm
all for shorter release cycles, hopefully we will be able to return to
3-4 full releases a year after this.


>  If this is not a release why are we doing test builds, aren't the builds
>  on graphical enough?

Unfortunately not, the graphical builds result in only a small amount
of usage and bug testing, bug reporting rate jumps drastically with
official test builds.

>  Lets skip this monkeying around, when the peach
>  team says ok builders were tagging please build.

Well the current plan is a release by between mid April and the end of
April, that will give time to fix the remaining issues that are felt
to be 'critical'.  Brecht said it will take roughly two weeks to get
the point cache issue ready to go.  Then there is also the work of
getting the rest of the particle system back up to full functionality,
and getting the UV editing back to its 2.50 functionality.

Once the bug tracker either 'hits 50' or gets down to what is agreed
to be an acceptable level then we can do a release.

LetterRip


More information about the Bf-committers mailing list