>
> A question: How do you go about creating the 'beta release' that is
> placed into the 'group file section'?   Do you start with a fresh
> 'snapshot' of the whole SVN trunk?


of course


>  If so, please add a SVN tag so we
> know exactly what you started with.
>

Please explain why you want to use a tag ? Isn't just a revision number
enough ?


>
> Then, I am proposing a testing/waiting period. Participation in the
> testing is voluntary.  You feel that 30 days is too long, so how about
> 21 days?
>

Yes, no, why not, I don't know :) I think there's no definite number of day
(maybe 42...), it depends on available people at that moment and release
content itself. Also, and I insist, we're using continuous integration with
buildbot, TORELEASE, weekly packages, unittests, etc... This is not a
big-bang release, releasing in our case means stating few things (changelog,
branch) and that's it, because the differences between a weekly packages and
a final one are mininal.

Again, please explain what you're testing during this "testing period". And
what will occur if there's not enough volunteer ? How will you track what
you've tested and what you haven't. With which jallib version ? With which
compiler version ? With which PIC ? You should also search the group for
"testing matrix" or things related to how to test jallib (hours of reading I
guess). Also google for "good enough software"...

AFAIC, I only trust automated tests, like unittests. Ideally, ultimately,
we'd need some kind of functional compiler specifications we could write
unittests against. That way you would be able to build a safety net, to test
new compiler versions, and automated as least in PC-in-silico. I try to do
this, see this unittest about new "alias" keyword:
http://code.google.com/p/jallib/source/browse/trunk/test/unittest/jalv2_alias.jalt.
It may someday detect a problem... or not. We'd need the same for some
jallib libraries. I wrote a bunch of tests for ADC libs, and they occur to
detect a problem in device files (and also fix a highly complex environment,
with many, many variables).

Finally, the fact is if there's a problem, we're able to fix it fast by
tracking changes with SVN, and release a new version also fast, thanks to
our continuous tools. Not every project can do this, that's also a safety
net.


>
> Now then, when we get to the end of the 21 days, and assuming we want
> to make this release 'official', the question comes up-- do we release
> exactly the same files as the 'beta release', or do we get a new
> snapshot from trunk, and generate a new SVN tag?
>

As I said, I used to create branch for beta, then another for final release.
By experience, it occured it wasn't that necessary. Keep in mind that things
don't move that much between beta and final. Particularly because we have
TORELEASE file acting as a kind of buffer. Between beta and final, people
can still commit new libraries/modications, if it doesn't hit TORELEASE,
that's transparent.

We need to be pragmatic. So far we haven't had any problems.

I understand you want to add tag (I still have my doubts about how useful it
is *in our case*), and specify a fixed number of days between beta and final
(I have doubts too about a fixed number of day, I hope you understood we're
already waiting days between beta and final). Is that right ?

Anyway, not a big deal...


Cheers,
Seb

--

You received this message because you are subscribed to the Google Groups 
"jallib" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/jallib?hl=en.


Reply via email to