Maven 1/2 support repo ?
Maven 1/2 support repo ?Guys I am a bit puzzled by how difficult is to maintain Maven (being
it 1 or 2) support for TestNG. Also, during the last conferences I
took part, I heard many people complaining that they are not able to
move to TestNG because of the lack/weak support it has in Maven.
Unfortunately, I am not a Maven user, nor do I have Maven architecture
knowledge. But, I would definitely like to hear for people with more
insight:
- what are the problems they are facing
- how can we have better support in Maven 1/2
- how can we make it more stable
Please try to answer these questions, and once we identify the
problems than I guess we can start addressing them one by one. Also, I
think we can ping some Maven guys for help if needed. Also, feel free
to point any other problems that are not covered by the above
questions.
think that your intentions are good but I'm unsure about a couple points.
-) Reporting of issues - Maven2 has a JIRA instance up already for surefire
bugs / etc. It isn't owned by us but I've been using it successfully so far.
-) Ownership of code - As much as I am happy to be able to use TestNG in
certain maven2 based projects I'm not sure we would want to formally
condone/embrace the cluster fuck that this has sometimes turned into. If we
truly had control over the code then that would be one thing, but hosting a
competing set of surefire plugins that people need to use a special testng
repo for sounds like a can of worms that would be better left unopened.
-) Status of fixes - All of the surefire fixes have been done/completed
already. There is a new committer on the maven2 dev team that has stated
he'll take patches from us and try to commit them ASAP as he also uses
TestNG and would like everything to be working smoothly.
-) Changing repos again - Enough people are already annoyed - but have
begrudgingly gone through the trouble of adding the
howardlewisship.comrepo. What do we tell them when we say "now use the
testng repo" ? I'm sure
some of them might think "why don't you just fix the f#%#-ing bugs instead
of creating new repos to host this mess in? " I would be one of them.
I think it will take the same amount of effort to try and create a new repo
host as it would for me to just create the final set of patches and have
maven2 incorporate the latest/greatest. Which would we rather have? Problem
gone and soon to be solved or move of cluster f#$ to a repo even more
closely associated with TestNG ? (which might also unfortunately start to
make TestNG look unstable if we are affected by code outside of our
control...I don't think thats fair to any of you. )
Of course if you are only talking about the ibiblio releases of official
TestNG jars then that is another matter entirely. I would fully support any
efforts in that direction, though I'd say it would be critical for us to get
an agreement setup with ibiblio to have the repo synced with them so that
people don't need to add any new repositories.
Those are my thoughts at least, if you'd still like to go the way you
outlined I'll "play ball" even if I don't agree.