FYI as of March 15th (10 days now), commits to branches following the
pattern (lucene|solr).* (i.e. that which start with "lucene" or "solr")
will *not* get an automated comment on corresponding JIRA issues. All
others continue to. INFRA got this done for us:
+1 I'll do it.
On Wed, Feb 3, 2016 at 3:29 PM Mark Miller wrote:
> Yeah man, ugly noise. I think we should open a JIRA to discuss with INFRA
> anyway. At a minimum, we should be able to specify which main branches we
> care about. Creating noise for everyone's random work
https://issues.apache.org/jira/browse/INFRA-11198
I don't think this has to do with some branches being more significant than
others. I think it has to do with the fact that the commit has occurred
multiple times (same hash); we just care about the first chronologically.
Any way; further
Thanks for creating the issue David. I did not mean to suggest we shouldn't
try, only that I think it will be an uphill battle. But i would be very
happy if something comes of it!
On Wed, Feb 3, 2016 at 1:39 PM, Mark Miller wrote:
> It just depends on what behavior you
It just depends on what behavior you want. Some could argue that for
certain 'official feature branches' they want this behavior and for random
WIP branches they don't. Given the script INFRA uses is for all the
projects, if they object to letting us only get the original commits, it's
just
Yeah man, ugly noise. I think we should open a JIRA to discuss with INFRA
anyway. At a minimum, we should be able to specify which main branches we
care about. Creating noise for everyone's random work branches is crazy.
- Mark
On Wed, Feb 3, 2016 at 3:09 PM david.w.smi...@gmail.com <
: https://issues.apache.org/jira/browse/INFRA-11198
: I don't think this has to do with some branches being more significant than
: others. I think it has to do with the fact that the commit has occurred
: multiple times (same hash); we just care about the first chronologically.
: Any way;
> unless i'm missing something, only getting email/jira
> notifications the first time a specific commit was seen would mean that
> when backporting from master to 5x (or 6x down the road) there would be no
> tracking email/comment ... which would make it much harder to know when
> things were
Right; I can't imagine any of us actually trying to merge master to
branch_whatever. We cherry-pick. Nobody has dissented on that to my
knowledge.
~ David
On Wed, Feb 3, 2016 at 8:19 PM Ryan Ernst wrote:
> > Are you sure *all* back porting will be done by cherry picking?
>
>
You can just as easily merge form another branch to master and we want that
commit message.
Also, there are certainly dissenters of cherry picking, a little Google
shows the arguments. Some do not believe in cherry picking at all, it's
certainly not a required operation.
If anyone got consensus
: > unless i'm missing something, only getting email/jira
: > notifications the first time a specific commit was seen would mean that
: > when backporting from master to 5x (or 6x down the road) there would be no
: > tracking email/comment ... which would make it much harder to know when
: >
> Are you sure *all* back porting will be done by cherry picking?
Using merge commits for backporting would require resolving all differences
between master and the stable branch. From what I've seen, using merge
commits between two long lived branches usually happens in other projects
by forward
There was 0 consensus from my perspective as well. Some projects make a new
branch and merge changes from it. You can do so many myriad of things,
which is why most projects come to guidelines and agreement on an approach,
but we seem to have some contributors of the mindset that if the new tool
: by forward porting, not backporting. I thought this was pointed out (that
: cherry picking is really the only way to backport) in an earlier git
: thread, but I could be wrong.
I will take your word for it.
(I ignored most of the early "if/when/should/why" threads about switching
to git
Ok thanks. It's too bad it's not going to be changed. IMO, I don't these
extra commit notifications are helpful beyond the original appearance of a
given commit.
~ David
On Tue, Feb 2, 2016 at 10:52 PM Ryan Ernst wrote:
> By the way, I meant I attended a talk *about git at
By the way, I meant I attended a talk *about git at apache*
On Feb 2, 2016 19:50, "Ryan Ernst" wrote:
> This is expected, as merging in the commit to a branch does create a new
> commit for that branch (or rather it is then visible in the history of that
> branch while before it
This is expected, as merging in the commit to a branch does create a new
commit for that branch (or rather it is then visible in the history of that
branch while before it was not).
Two years ago at ApacheCon I attended a talk by David Nalley. This issue
was brought up, and I spoke a little with
17 matches
Mail list logo