Hi Miroslav,
On Fri, Apr 08, 2022 at 11:01:49AM +0200, Miroslav Zagorac wrote:
> Hello Willy,
>
> the attachment contains patches for haproxy branches 2.4 and 2.[56].
>
> Branch 2.4 does not have several patches because they are not
> applicable to it. The numbering has stayed the same, so you can
> handle it easier.
Thanks, but now I'm having further questions:
- 0006-BUG-MINOR-opentracing-setting-the-return-value-in-fu.patch
- 0016-BUG-BUILD-opentracing-fixed-OT_DEFINE-variable-setti.patch
=> these ones are bug fixes that are independent on the series, and
I verified that they could apply fine before the rest. Do you
mind if I do it ? We always want to include fixes before the
rest so that bisection is more accurate.
- 0013-EXAMPLES-opentracing-refined-shell-scripts-for-testi.patch
=> same for this one, I'd put it just after the two fixes above.
- 0008-DOC-opentracing-corrected-comments-in-function-descr.patch
=> this one, among other things, fixes the description of function
flt_ot_smp_init() that was added as part of the series, so at
last this part of the fix should be merged with the patch that
introduced the function (no need to integrate an incorrect
patch and fix it later), and the rest are fixes for the past
so I could put it earlier in the stack, with the fixes.
- 0014-MAJOR-opentracing-reenable-usage-of-vars-to-transmit.patch
=> while I'm fine for 2.6, I'm really not for 2.5 without a big
compelling reason. It's a feature addition, not a bug fix.
Either there are currently no opentracing users in 2.5 due to
this problem and this will serve nobody, or there are and there
is a significant risk of breaking something for them, which would
possibly encourage them to revert an update and keep other unrelated
bugs. Thus is it absolutely mandatory that we backport this to 2.5 ?
For example, is it currently working so bad that users complain and
that it prevents them from upgrading from 2.4 to 2.5 ? That's the
type of situation that can justify a backport of such a patch.
Please do not send me another series and just let me know your opinion
about the points above, I'd like to issue 2.6-dev5 right now, ideally
with these patches, and do not want to iterate through yet another
series.
Regarding the 2.4 series, I'm seeing 5 cleanup patches, but I haven't
checked if they were needed or not. Normally we do not backport code
cleanups in stable branches unless they are tiny, riskless and help to
get other patches backported by providing the required patch context.
One of the reasons is that some users have patches on top of these
branches and applying cleanups that are not necessary sometimes causes
them trouble to re-apply their local patches. It's unlikely that users
have patched opentracing but that's a general rule, I think you get the
idea.
Thank you!
Willy
