On 08/07/2008, at 3:28 AM, Igor Fedorenko wrote:
Brett,
Out of curiosity. It seems that your fix for MARTIFACT-25 only
caches missing pom.xml lookup. Where is the logic that will prevent
maven from repeatedly hitting remote repositories for actual
artifacts?
I'd looked at the linked issues which referred to servicemix and
missing POMs, and I thought that was what John was referring to as well.
I hadn't originally thought of the case of optional dependencies
possibly not being resolvable - that's not what I meant for optional
to mean :) Regardless, they should only be requested once in a
resolution cycle for artifacts.
So I'm not quite sure how you're getting this. Can you reopen the
issue, and attach a test project that illustrates multiple resolution
failures?
At this point, I don't believe that caching this in the artifact
resolver is the right place. The simple potential for failure such as
the following concerns me:
- request a:1.1 (404)
- deploy a:1.1
- request a:1.1 (fails, despite being present)
Even doing this for POMs is somewhat questionable, but I figured an
acceptable trade-off given the common existence of artifacts without
POMs vs the likelihood of this happening (and ease of correcting it
with -U).
HTH,
Brett
Brett Porter wrote:
Hi Oleg,
I think you are good to go now for the release. I've fixed the
additional issues that I'd found.
I'm just running the integration tests again locally to confirm,
but I'd say go for it whenever you're ready. All the instructions
are on the site for setting up settings, signatures, etc. if you
need it, though I'm sure you have it covered :)
Cheers,
Brett
On 07/07/2008, at 4:50 AM, Brett Porter wrote:
On 05/07/2008, at 1:51 AM, Oleg Gusakov wrote:
John - are you looking into this? Need it resolved for MNG-3185
I've done it in the way John was suggesting - it's not the perfect
solution but it's the best one available I think.
I stumbled upon an API regression that breaks one of servicemix's
plugins in the process (MARTIFACT-27). I'm not going to make it to
the end of this tennis match, so I'm going to bed for now :)
If there aren't any other takers on this I can take a look
tomorrow. I think it'd be worth running clirr on this too - but
it'll require some fiddling since the artifact has changed and
been merged.
Cheers,
Brett
Thanks,
Oleg
Brett Porter wrote:
On 04/07/2008, at 3:40 AM, John Casey wrote:
Not to distract from the higher-level discussion, but I'd like
to get into the nuts and bolts of MARTIFACT-25 a bit...in case
someone beats me to it.
We introduced a properties file that tracks resolution attempts
for artifacts that weren't found on the remote repository, and
I'd like to see if we can reuse that file/concept/code to
handle artifacts that don't have accompanying POMs on the
remote repo. It's a similar concept, so the two should dovetail
relatively well, and require little code to accomplish the fix.
This makes sense to me. So continue to expand on the update
check manager to handle this case?
The other alternative - which at one point we were doing - is to
write out the stub POM into the local repository and retain
that. It would simplify the code, but muddies the local repo
content.
I believe these have the same net effect, in that the artifact
is never resolved a second time. Is that correct?
The way I see it, the biggest hurdle for this issue is creating
a [set of] good tests to circumscribe the issue and make sure
it doesn't regress later.
I don't think there's too many variations - the problem is only
with missing POM files, right?
Cheers,
Brett
My $0.02.
-john
Brett Porter wrote:
Hi all,
As I indicated a couple of weeks back, moving towards a 2.1
alpha release, I was looking at releasing an alpha of maven-
artifact. Brian was able to locate the issue he was referring
to a couple of weeks back about re-resolving (now
MARTIFACT-25), so I've postponed.
Once that's fixed, I think there should be no reason not to
proceed with a release based on the thread we had back then.
Are there any other issues that anyone sees as blocking moving
forward with a release?
I've done all the testing I was planning to already. Oleg, did
you want to take this release, or shall I go ahead after that
issue is sorted?
Cheers,
Brett
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
--
Brett Porter
[EMAIL PROTECTED]
http://blogs.exist.com/bporter/
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]