[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
[gemini-dev] DBAccess: Status and Plan
|
Here is an update of the status of DBAccess:
With the completed Graduation and 1.0.0 Release Review we have cleared the way for the 1.0.0 release (http://dev.eclipse.org/mhonarc/lists/gemini-dev/pdfvFVchVdSys.pdf).
However, according to an issue with "Eclipse linked resources", the MySQL contribution (and refactoring of the project) was neither part of the Release Review document, nor of the head revision (see http://wiki.eclipse.org/Gemini/Meetings/Minutes_Feb_8_2011 ).
I would like your opinions on the following items:
1) I guess, It would be desirable to have the DBAccess release close to the Gemini JPA release (this month, next month?)
2) I have therefore branched off a 1.0.0 in svn.
3) Unless the absence of MySQL really hurts, I would prefer to create a release candidate from the current branched version (technically, I do not know, if including MySQL would require another Release review).
4) We have two independent rather big contributions to DBAccess via Bugzilla (from Krum Tsvetkov and David Schmitz). Both address the continuous integration process with Maven and Tycho, and David also expands the supported database drivers.
5) I think automated builds via Maven/Tycho are a good thing to have, and I would be in favor it.
6) Based on the Krum's and David's work, I experimented with Maven/Tycho: Technically, it is tricky to support Eclipse workspace dependencies with Maven/Tycho. So the dreaded "Eclipse linked resources" that already caused some headache are definitely not supported by Maven/Tycho. My suggestion would be to start with a separate bundle for the common part. If that split turns out to be too awkward, there is way incorporate it into each of the other bundles (I have checked that). However, my Tycho solution for that would not be very elegant. Additionally, there is always the danger of OSGi ClassCastExceptions when replicating packages into different bundles.
7) My suggestion would be to start with a simple Maven/Tycho based project setup that would surely not contain all of David's and Krum's suggestions.
8) Starting from that, it should be possible to progress the project in an incremental way.
9) Especially Gerrit (Git) would be nice for integrating these smaller increments, but I am not sure if the advantage of Gerrit over ordinary diffs would justify the migration effort at this stage.
10) I am not exactly sure about the release time frame of that "2.0". Any suggestions, comments? Any requirements(from JPA)? Clearly, if we manage to setup an automated build, it would be easier for others to consume intermediate snapshots...
-Juergen
