Community
Participate
Working Groups
The current PDE build consume binaries checked into the git repo and creates all of the launcher fragments to be used by the build p2 repo. CBI platform build needs to be able to correctly build the launcher fragments and include them in the published build p2 repo. PW
Is this bug the same as Bug 378233?
I thought this was allowing a CBI build to build the launcher and its native fragments ... but on one machine, this leads to a "slice" repo 'cause it can only build for the platform we're on. Bug 378233 is so that CBI build can consume the launcher binary repo and generate a p2 repo that has all of the different platform fragments (more like today's PDE build) PW
Is this bug about building eclipse.exe for Windows, and the eclipse natives on Mac and Linux?
See also bug 405753 and bug 415757 which I think are related (if not dups?)
Building natives on linux is a matter of passing -Dnatives=gtk,linux,$your_arch to mvn. If you have all headers installed in the expected places you will get your natives rebuild and used.
(In reply to David Williams from comment #4) > See also bug 405753 I would certainly recommend closing bug 405753 as a dupe of this one. Building the launchers would be done as part of the CBI project.
(In reply to Denis Roy from comment #6) > (In reply to David Williams from comment #4) > > See also bug 405753 > > I would certainly recommend closing bug 405753 as a dupe of this one. > Building the launchers would be done as part of the CBI project. Yes, and no, IMHO. I could see it either way. But, if you define "CBI" as the infrastructure to do "common build" things, then there is a piece that CBI would have to do -- to provide the capability to make it easy, useable by others, etc. ... but, the responsibility for actually doing it would still reside with the Platform releng (or swt :) team. There are other projects, BIRT? CDT? ... maybe others ... that have some about of native code. I'd assume that "CBI" would provide the infrastructure for anyone to "build natives" (on the common platforms). Right? Or, am I missing the point.
(In reply to David Williams from comment #7) > (In reply to Denis Roy from comment #6) > > (In reply to David Williams from comment #4) > > > See also bug 405753 > > > > I would certainly recommend closing bug 405753 as a dupe of this one. > > Building the launchers would be done as part of the CBI project. > > Yes, and no, IMHO. I could see it either way. But, if you define "CBI" as > the infrastructure to do "common build" things, then there is a piece that > CBI would have to do -- to provide the capability to make it easy, useable > by others, etc. ... but, the responsibility for actually doing it would > still reside with the Platform releng (or swt :) team. > > There are other projects, BIRT? CDT? ... maybe others ... that have some > about of native code. I'd assume that "CBI" would provide the infrastructure > for anyone to "build natives" (on the common platforms). > To capture this relationship/idea, I've changed title of 'releng' bug to Bug 405753 - Make use of CBI to build natives on Eclipse Foundation instrastructure and marked it as "depends on" on this one. I'm actually not sure "where the line is" ... but, am sure it will take "both of us" to do something ;)
> but, the responsibility for actually doing it would > still reside with the Platform releng (or swt :) team. Platform releng, Birt and anyone else can build the natives if they want. I want to be able to build them too, perhaps so we can enhance them for usage in the EPP packages. My end game is bug 445141. That is what this bug is about.
I've configured the job to run on the CBI CentOS slave.. It still fails miserably, but now we have a more modern platform we can work with. https://hudson.eclipse.org/cbi/job/cbi-launcher-linux-x86_64/11/console
(In reply to Denis Roy from comment #10) > I've configured the job to run on the CBI CentOS slave.. It still fails > miserably, but now we have a more modern platform we can work with. > > https://hudson.eclipse.org/cbi/job/cbi-launcher-linux-x86_64/11/console There is smth wrong with the slave. I switched it back to master and played a bit with it and voilla https://hudson.eclipse.org/cbi/job/cbi-launcher-linux-x86_64/22/ :). Let's find out what's wrong with the slave.
Next step should be to make it buildable without rt.equinox.binaries so we can be 100% sure that a prebuild artifact doesn't sneak instead of the just build one.
cc'ing Mikael
native are now build on releng's jipp. All test jobs on cbi jipp have been removed.