[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [tycho-dev] FW: using <feature-id>.feature.group as artifactId

See inline...

--
Regards,
Igor

On 11-11-29 6:19 AM, Oberlies, Tobias wrote:
Opening to all tycho developers...

-----Original Message----- From:
ldi-developers-bounces@xxxxxxxxxxxxxxxxx
[mailto:ldi-developers-bounces@xxxxxxxxxxxxxxxxx] On Behalf Of
Oberlies, Tobias Sent: Dienstag, 29. November 2011 12:14 To: Sievers,
Jan; Igor Fedorenko Subject: Re: [ldi-developers]
using<feature-id>.feature.group as artifactId

I am -0 on supporting artifactId=<featureId>.feature.group. AFAIK the
".feature.group" suffix is not API of p2, so we shouldn't make it API
of Tycho. But I see that this could be an alternative workaround to
the GAV uniqueness problem.


For all intents and purposes artifactId is internal to Tycho, so I don't see a problem here.

I am -1 on disallowing artifactId=<featureId>. This would mean that
people who name their features "org.example.tool.feature" would be
forced to set the groupId "org.example.tool.feature.feature.group".
This makes it easy to use the same IDs for features and bundles
(which is IMHO a mediocre idea), and put burden on people who name
their features properly.


Fair enough.


In any case, Tycho would need to cope with a non-unique mapping
between artifactId and Eclipse Id (at least in the transition phase)
- and if we need to implement this anyway, we could also just keep
it. One reason for not allowing arbitrary artifactIds is that fixing
all the problems it entails [1] was deemed too much effort.


Supporting two variants is far easier than allowing arbitrary feature-id/artifacyId mapping.



Regards Tobias


[1] https://bugs.eclipse.org/bugs/show_bug.cgi?id=353384


-----Original Message-----
From: Sievers, Jan
Sent: Dienstag, 29. November 2011 09:45
To: Igor Fedorenko; Oberlies, Tobias
Subject: RE: using<feature-id>.feature.group as artifactId

+1

the case where there are plugins and features with the same id is
unfortunately pretty common and our workaround with a different groupId is
ugly.

Appending .feature.group is consistent with how p2 solved the same
problem.

Regards
Jan

-----Original Message-----
From: Igor Fedorenko [mailto:igor@xxxxxxxxxxxxxx]
Sent: Montag, 28. November 2011 21:00
To: Oberlies, Tobias; Sievers, Jan
Subject: using<feature-id>.feature.group as artifactId

One thing I forgot to mention today.

To disambiguate features and bundles with the same id, p2 adds
.feature.group suffix when generating IU id from feature id, if feature
id does not end with .feature.group already.

I would like to do the same in Tycho, only do this more aggressively.
For 0.14 allow artifactId with and without .feature.group suffix but
when .feature.group is not present. For the next version only allow
artifactId=<feature-id>.feature.group


In other words, 0.14

feature.xml id=org.some.feature
pom.xml artifactId=org.some.feature =>  WARNING
pom.xml artifactId=org.some.feature.feature.group OK


In the next version

feature.xml id=org.some.feature
pom.xml artifactId=org.some.feature =>  FAILURE
pom.xml artifactId=org.some.feature.feature.group OK


I really dislike using special groupId and I think there are cases where assuming unique artifactId for entire reactor build will make things easier.

What do you think?

--
Regards,
Igor
_______________________________________________
tycho-dev mailing list
tycho-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/tycho-dev