|
On 03/23/2015 02:34 PM, Mickael Istria
wrote:
TL;DR: Being in master != being in
delivery for Mars.
On 03/23/2015 12:21 PM, Victor V. Rubezhny wrote:
I didn't merged the code in master,
since we still have few issues that blocks us from doing it in
Mars (I don't think we can get them fixed in time for Mars, so
it looks like we're missing Mars stream with this
contribution).
That's not a problem: we can have code in master without shipping
it on Mars, that's what I strongly believe is the best approach
for easier collaboration, and to avoid risk of diverging branches.
IMHO, It's better to write your
suggestions here (in mailing list) first, because I'm going to
push another patch set into Gerrit ( https://git.eclipse.org/r/#/c/41693/
), so if you have some code comments to do then your comments
could be 'forgotten'. But still free to add comments to the
change.
Suggestion is that for a 1st integration in master, just include
the bundles, not the pom files, do not update other pom files,
feature.xml nor category.xml. So the code is there, people can
work on it, but not altering the build and delivery.
But this will not enable neither the building of Bower plug-ins, nor
running its JUnit tests then.
Then
next iteration would be to see if we can build it easily, include
in the category.xml file as a separete entry, without modifying
the "regular" JSDT feature so it wouldn't get in Mars. That would
allow people who want to try it to get it from a p2 repository.
I can create a category for Bower IDE, but again: what will go into
this category in case we aren't adding the Bower plug-ins into the
build? It looks for me that we have to add the bower-related
plug-ins to pom.xml-s, to create a separated feature for bower and
then include that feature into site/pom.xml and category.xml, but
not to jsdt.feature/jsdt_sdk.feature. But I'm not sure if bower
plug-ins will not became 'installable by default' in such case.
IMHO, contributing it into master
will not make any harm (visually it will add few actions to
context menus) for WTP and JSDT, all the 'issues' are only on
how Bower IDE works. But it looks unfinished at the moment and
still have few important issues to be fixed before that
possible merge to master. M6 is going to be declared today or
in few days, so it looks like it's too late to merge. I'm sure
we can finish it for the Mars+1 stream.
But I'd still like to hear/read on your suggestions.
Being in master and being in Mars are totally separated topics,
both technically and "organisationally". The most efficient and
easier approach would be to have code in master right now, but not
ship it in Mars. The fact that M6 is approaching doesn't forbid
merging unrelated changes. It's actually something we could even
add on the day before Mars if we wanted ;)
_______________________________________________
wtp-dev mailing list
wtp-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/wtp-dev
|
