A quickly answer:

On Sat, Jan 23, 2010 at 1:45 AM, Viktor Szakáts <[email protected]> wrote:
> Hi,
>
>> 1) hbtrace.c
>> only some changes to hbtrace are marked TOMERGE and so the marked
>> patches don't apply.
>> Can I use the TRUNK version ?
>
> I think you can't. Several features were added
> to this component, so it doesn't count as pure
> bugfix. To avoid regressions, no new features
> should be back-ported to 2.0.x as a rule.

Ok

>
>> 2) hbmk2.pt_BR.po
>> there are several commits for this file and for some of them the label
>> [TOMERGE] was written and then removed... PLEASE don't do it again. If
>> a file "file.prg" has a "complicated" history due to test with poor
>> results there are two possibilities: keep the [TOMERGE] label on every
>> revision on clearly say in the commit "take directly this revision
>> dropping the history" actually collapsing several commits into one.
>> Can I use the TRUNK version ?
>
> Please you the version committed in:
>   2009-12-23 02:59 UTC+0100 Viktor Szakats (harbour.01 syenar.hu)
>
> (as for 'don't do it again': you can safely assume
> that such commits are only done when there is some
> serious issues along the way, in this case it was
> codepage problems came at last minute before release.
> That's life, such thing can happen in development ,
> unfortunately. Anyway, multiple modification should
> not pose a problem in general.)

It's ok, I just wanted to say that for clean patch apply all the
commits should be applied....

>
>> 3) 2010-01-05 18:48 UTC+0100 Viktor Szakats
>> only in hbmk2.pt_BR.po, probably due to not applied patches at point 2.
>
> I'm not sure what you mean, but this specific
> patch of this file shouldn't be merged.
>
> The hbmk2.prg -warn fix should go though, it's definitely
> a manual merge, since multiple changes were done in
> this one commit.

I will check later.

>
>> 4) 2009-12-31 12:43 UTC+0100 Przemyslaw Czerpak
>> it doesn't apply, but I need to investigate better (probably due to
>> some missing previous codepage patches)
>> Should all codepage rfelated patches be MERGED ?
>
> This should go as is. hbext*.ch files may need
> to be applied manually.
>
>> 5) 2010-01-13 20:14 UTC+0100
>> it doesn't apply because it must be applied to lines added by
>> "2010-01-13 09:37 UTC+0100 Przemyslaw Czerpak" that was not marked as
>> TOMERGE that added some similar lines in the same lines confusing the
>> patch system. Should I MERGE it ?
>
> This change is not marked as TOMERGE, so you shouldn't.

But 2010-01-13 20:14 UTC+0100 adds a line in function
hb_hashGetCItemPtr() added in "2010-01-13 09:37".... or we add both or
none...

>
>> 6) 2010-01-18 13:27 UTC+0100
>> mapi.c doesn't apply for different formatting.... no problem
>
> I can't see the exact problem, but the fix it rightly
> marked as TOMERGE, maybe it needs to be manually applied.
>
>> PLEASE PLEASE PLEASE, in order to not destroy the work I have done up
>> to now, please DON'T make changes to ChangeLog but instead tell me in
>> this message what should I do !
>>
>> 29 patches apply cleanly.
>
> Thank you. My only comment is that nobody ever told
> that patches should or would apply cleanly :( There are
> cases when things has to be done manually. IMO it's
> almost impossible to design daily "life" around making
> back-porting patches a no-brainer. [I've went through
> it in 1.0.1, which I did fully manually BTW.]

I'm doing it quite manually to extract commits and hunks.

But you are a "core" developer, I'm just helping and I can't know if a
patch is really a bug fix or something else...

For example in one of your patches (about unicode overflow protection)
there were changes to win_prn3  (TEXT(0)) not listed in the Changelog
and I can't know if they are needed or not... in anay case, they don't
apply..


>
> Based on merging experiences we may try to create some
> basic committing rules to help the process though, but
> IMO there is no way to _fully_ avoid automatic merge issues.

Yes, for example if you set a [TOMERGE 2.0] label for previous
commits, the comit message should be
Added [TOMERGE 2.0] to rev XY
and not "updated previous Changelog...."


Francesco
_______________________________________________
Harbour mailing list (attachment size limit: 40KB)
[email protected]
http://lists.harbour-project.org/mailman/listinfo/harbour

Reply via email to