[1.0-M3] released .

[Laurent Forêt]

Hello every body,

I just have taged the sources a

[Laurent Forêt]

Hello everybody,

I just have tagged the sources and put a downloadable release for the Milestone 3.

The new features are :

1. settings.xml file code completion support.
2. handling ivy Configuration (mapping configuration with netbeans compile, test, runtime scopes).
3. resolving and downloading of artifact sources and javadocs
4. resolve optimization with last resolve reuse
5. module publish support in build task.

The publish action is made while the jar task is called, it is working fine and we can manage project dependencies using ivy mecanisms.

To have ant build scripts workink fine with ivy the http://www.netbeans.org/issues/show_bug.cgi?id=138661 should be resolved.


Best regards.

Laurent.

[Milos Kleint]

hello

how is the code completion created? I've got it for the org quite
fast, but the name and revision ones seemed to hang forever for me..
both in ui and editor..
when I add commons-httpclient and then use digestutils from
commons-codec, I make it work in the editor, but it fails when
building the project, I suppose it's
http://www.netbeans.org/issues/show_bug.cgi?id=138661 related, right?
is there a workaround or is it critical 138661 gets done?

apart from that, the code completion in java files works now, great.
Ivy support will even download javadoc and sources, cool.

regards

Milos

[Laurent Forêt]

Inline

On Mon, Jul 28, 2008 at 9:37 AM, Milos Kleint <mkl...@gmail.com> wrote:

hello

how is the code completion created?

Xavier will answer better than me for that point. I was also supprised by the fast listing of orgs.
 

I've got it for the org quite
fast, but the name and revision ones seemed to hang forever for me..
both in ui and editor..

Yes, I reproduce it to, It has worked fine before, I think it is a regression.
Xavier, what do you think about it ?
 

when I add commons-httpclient and then use digestutils from
commons-codec, I make it work in the editor, but it fails when
building the project, I suppose it's
http://www.netbeans.org/issues/show_bug.cgi?id=138661 related, right?
is there a workaround or is it critical 138661 gets done?

Yes, the work around is : to launch all ivy tasks (resolve and retrieve ) in the init phase of ant scripts. It should work (has worked before) but even if you do a clean task you will launch the retrieve task, that is stupid.

I would rather that the issue 138661 was (what is the "subjonctif "in english ?)  resolved, but I will change the ivybeans code accordingly to the status of this issue before the 4/08 .

However, the patch is very simple: the idea is to add the "classpath" attribute to all the javac macro.
 

apart from that, the code completion in java files works now, great.

Yes !!!
 

Ivy support will even download javadoc and sources, cool.

Thanks and in a transparent manner, I like that. Differently from the maven plugin, you can browse the sources from the library node if sources exist, it is nothing to do but it is nice.


Regards,

Laurent.
 

[Milos Kleint]

ahuradha (from mevenide team) tested the module as well, here's the
exception he got (told me to pass on you)
http://statistics.netbeans.org/analytics/detail.do?id=82978

Milos

[Laurent Forêt]

He found the time to do this ;) !

It is a problem of retrieving libraries from repositories, I should do a better handling of that functional exceptions.
I think that I should use the same mecanism from mevenide to notify that a library is missing :the minus badge on that library.

I add http://code.google.com/p/ivybeans/issues/detail?id=1.

I will have a look at, this evening.

Laurent.

[Xavier Hanin]

On Mon, Jul 28, 2008 at 9:37 AM, Milos Kleint <mkl...@gmail.com> wrote:

hello

how is the code completion created? I've got it for the org quite
fast, but the name and revision ones seemed to hang forever for me..
both in ui and editor..

Org, module and revision completion relies on Ivy listing feature, which depends on the underlying settings, and how the repos responsiveness. For  the org, Ivy maven 2 repo implementation does not provide org listing at all, so you only get org listing from your local resolvers, which works very fast. Then for module and revision listing, Ivy does an http GET on the server to get the listing of a directory, and depending on your internet connection and the server responsiveness, it can be very long. This could be improved with caching, and /or indexing (with nexus for instance). But I think Ivy is currently used most of the time with in house repositories which provide very good response time. So maybe we can delay this improvement for after the 1.0.

Xavier
 

--
Xavier Hanin - Independent Java Consultant
http://xhab.blogspot.com/
http://ant.apache.org/ivy/
http://www.xoocode.org/

[Milos Kleint]

On Mon, Jul 28, 2008 at 6:02 PM, Xavier Hanin <xavier...@gmail.com> wrote:
> On Mon, Jul 28, 2008 at 9:37 AM, Milos Kleint <mkl...@gmail.com> wrote:
>>
>> hello
>>
>> how is the code completion created? I've got it for the org quite
>> fast, but the name and revision ones seemed to hang forever for me..
>> both in ui and editor..
>
> Org, module and revision completion relies on Ivy listing feature, which
> depends on the underlying settings, and how the repos responsiveness. For
> the org, Ivy maven 2 repo implementation does not provide org listing at
> all, so you only get org listing from your local resolvers, which works very
> fast. Then for module and revision listing, Ivy does an http GET on the
> server to get the listing of a directory, and depending on your internet
> connection and the server responsiveness, it can be very long. This could be
> improved with caching, and /or indexing (with nexus for instance). But I
> think Ivy is currently used most of the time with in house repositories
> which provide very good response time. So maybe we can delay this
> improvement for after the 1.0.

sure,
in 6.5+ timeframe we could probably reuse what was done in the
maven+nexus workspace and mevenide. As the maven modules are moving
into standard distribution after 6.5, the nexus indexer codebase could
be easily accessible to ivybeans as well.

Milos

[Xavier Hanin]

On Mon, Jul 28, 2008 at 7:00 PM, Milos Kleint <mkl...@gmail.com> wrote:

On Mon, Jul 28, 2008 at 6:02 PM, Xavier Hanin <xavier...@gmail.com> wrote:
> On Mon, Jul 28, 2008 at 9:37 AM, Milos Kleint <mkl...@gmail.com> wrote:
>>
>> hello
>>
>> how is the code completion created? I've got it for the org quite
>> fast, but the name and revision ones seemed to hang forever for me..
>> both in ui and editor..
>
> Org, module and revision completion relies on Ivy listing feature, which
> depends on the underlying settings, and how the repos responsiveness. For
> the org, Ivy maven 2 repo implementation does not provide org listing at
> all, so you only get org listing from your local resolvers, which works very
> fast. Then for module and revision listing, Ivy does an http GET on the
> server to get the listing of a directory, and depending on your internet
> connection and the server responsiveness, it can be very long. This could be
> improved with caching, and /or indexing (with nexus for instance). But I
> think Ivy is currently used most of the time with in house repositories
> which provide very good response time. So maybe we can delay this
> improvement for after the 1.0.

sure,
in 6.5+ timeframe we could probably reuse what was done in the
maven+nexus workspace and mevenide. As the maven modules are moving
into standard distribution after 6.5, the nexus indexer codebase could
be easily accessible to ivybeans as well.

That would be great!

BTW, after more investigation the listing feature of Ivy was actually buggy in the version of Ivy we use d in Ivybeans. I've fixed the bug in Ivy and upgraded the version used in ivybeans, we shall have much better completion for module and revision in next release.

Xavier
 

[Laurent Forêt]

It is also fixed in trunk.

Laurent.

[Laurent Forêt]

Fixed to ! Unresolved dependencies are materialized by the ResourceNotIncluded.gif badge.

regards.

Laurent.