I agree with Ramana, and recently found a very nice tutorial describing different workflows using git branches:
https://www.atlassian.com/git/tutorials/comparing-workflows
I am using the "Feature Branch Workflow" in my open source projects, which allows me to commit and push even very rough pieces of code to a feature branch without breaking the trunk that other people might be using (e.g., Michael's Travis CI system).
Cheers,
Joe
On Tue, Mar 31, 2015 at 6:20 AM, Ramana Kumar <ramana at member.fsf.org> wrote:
I would be happy to see Poly/ML development happening in a git repository.
For the particular scenario you describe (various versions of the sources in stable and unstable states), I would just use different branches within one public repository rather than bothering with multiple public repositories. Branches in git are cheap.
On Tue, Mar 31, 2015 at 12:53 PM, David Matthews <David.Matthews at prolingua.co.uk> wrote:
This was a problem with the SVN version of Poly/ML. I've fixed it now so it should build.
This has raised, though, a question I've been wondering about: whether to switch from sourceforge SVN to git and github. It's very convenient to be able to commit works-in-progress at various stages even if they're not complete. It makes it much easier to track down bugs by bisection if each commit is not too large. The disadvantage with SVN is that it's not clear to anyone using the repository whether it is meant to be in a stable state and meant for beta testing or whether it is unstable and perhaps not even alpha tested. It's much easier with a multi-stage source management system such as git or mercurial to have a "private" repository and only push to the "main" repository when the set of commits are complete.
Does anyone have any thoughts on this?
David
On 31/03/2015 09:56, Michael Norrish wrote:
My Travis CI builds of HOL using a Subversion checkout of Poly (as opposed to the released version, which I also check), are failing to build Poly with the error below. (See also https://travis-ci.org/HOL-Theorem-Prover/HOL/jobs/56529081).
Is this expected, or some quirk of the Travis setup?
Michael
--
Error- in './basis/FinalPolyML.sml', line 1607. Value or constructor (decodeBinary) has not been declared Error- in './basis/FinalPolyML.sml', line 1608. Value or constructor (encodeBinary) has not been declared Exception trace for exception - Fail raised in mlsource/MLCompiler/MAKE_.ML line 423 MAKE_().useIntoEnv(2) <top level> CODETREE().genCode(3)(1) COMPILER_BODY().baseCompiler(3)executeCode(1) End of trace Exception trace for exception - Fail raised in mlsource/MLCompiler/MAKE_.ML line 423 MAKE_().useIntoEnv(2) <top level> CODETREE().genCode(3)(1) COMPILER_BODY().baseCompiler(3)executeCode(1) End of trace
polyml mailing list polyml at inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/polyml
polyml mailing list polyml at inf.ed.ac.uk http://lists.inf.ed.ac.uk/mailman/listinfo/polyml