On Sat, 17 Nov 2012, Jed Brown wrote: > On Sat, Nov 17, 2012 at 11:19 AM, Satish Balay <balay at mcs.anl.gov> wrote: > > > I guess it could b: > > > > - I make changes to petsc-release/buildsystem, commit > > > > This create a new head in petsc-release/buildsystem (but not in > petsc-release). > > > > push petsc-release > > [error with buildsystem push - so merge with petsc-release, push] > > > > - Barry makes changes to petsc-release/buildsystem, commit push > > petsc-release. > > Error with petsc-release push - so, pull, merge petsc-release. This will > > also pull,merge corresponding buildsystem. now commit and push > > petsc-release.
This presumably merged the new head [in petsc-release/buildsystem] that I created with the new head that Barry created in his clone? BTW: am I correct in assuming that this head won't get pushed to buildsystem unless 'petsc-release -> petsc-dev -> push to bitbucket' process is completed? So in that intermediate time - petsc-release will point to a non-existant buildsystem state? Or will a push of petsc-release won't succeed until the above 'petsc-release -> petsc-dev -> push to bitbucket' process is completed? > > [error with buildsystem push - so merge with petsc-release, push] > > > > Right, go to petsc-dev, pull petsc-release, merge (applies to both > petsc-dev and petsc-dev/buildsystem), and push. > > > > > > BTW: from http://mercurial.selenic.com/wiki/Subrepository > > > > * There's no support for merging across renaming/moving subrepos * > > > > So changing 'BuildSystem -> 'buildsystem' will cause grief with subrepos? > > > > I think we should do the rename first. Then we should perhaps wait till the next release [just before the release - when no more patches will be merged from petsc-3.3 to petsc-dev] before making these changes? [I don't think its appropriate to make this change for a petsc-3.3 patch update.] Satish
