Uploaded image for project: 'Request For Comments'
  1. Request For Comments
  2. RFC-546

adopt semver 2.0.0 for public releases of DM developed software

    XMLWordPrintable

    Details

    • Type: RFC
    • Status: Withdrawn
    • Resolution: Done
    • Component/s: DM
    • Labels:
      None

      Description

      At present, there is no documented policy on the format of version strings for DM software releases.  While researching the historical usage of version strings as part of improving the documentation of the science pipelines release process, it appears that (at least) 2-4 component version strings have been used in the past, presumably with varying semantic meaning for each field. Eg., lsst/afw has the tags 3.6.08.0.0.3 and 16.0 In addition, there isn't any rule that requires the versions strings to increment as sequence of positive integers. I suspect most DM devs would be uncomfortable with the next major release being versioned as 99.88, followed by -47.-32, followed by foo.bar.baz.

      In the interests of having a well-defined and non-ambiguous policy going forward I propose that semver 2.0.0 be adopted for versioning publicly released DM software. This policy would explicitly not apply to 3rd party software which is redistributed by DM nor DM producted software which is not formally versioned (utility scripts, etc.). Per package waivers to the usage of semver 2.0.0 would be requested via RFC.

      https://semver.org/spec/v2.0.0.html

        Attachments

          Issue Links

            Activity

            Hide
            gcomoretto Gabriele Comoretto added a comment -

            I am in favor of semantic versioning.

            However I think this will not be sufficient to solve the problems we have now in the release process.

             

            I am trying to understand and figure out what changes can be put in place in order to improve the release process, and semantic versions is of course one of these elements.

            I would appreciate some coordination,  with what I am trying to do, before opening RFCs that have an impact on the release process, and in general before implementing  changes on scripts and tools that also impacts the release process.

            Show
            gcomoretto Gabriele Comoretto added a comment - I am in favor of semantic versioning. However I think this will not be sufficient to solve the problems we have now in the release process.   I am trying to understand and figure out what changes can be put in place in order to improve the release process, and semantic versions is of course one of these elements. I would appreciate some coordination,  with what I am trying to do, before opening RFCs that have an impact on the release process, and in general before implementing  changes on scripts and tools that also impacts the release process.
            Hide
            jhoblitt Joshua Hoblitt added a comment -

            As there appear to be no objections, I'm going to mark this RFC as accepted.

            Show
            jhoblitt Joshua Hoblitt added a comment - As there appear to be no objections, I'm going to mark this RFC as accepted.
            Hide
            swinbank John Swinbank added a comment -

            Hi Josh — sorry, I didn't catch this one last week because I was travelling. I'd prefer that we walk this decision back for now for a few reasons:

            1. I'm concerned that references to version numbers exist in a number of baselined documents, and in the official project schedule. Changing the baseline can't be done without (at least) the approval of the DM-CCB and (possibly) the full project CCB.
            2. Per my comment on RFC-547. Gabriele Comoretto and Leanne Guy are in the process of drafting a policy about release versioning and support. I'd like to give them chance to make a proposal before we adopt any substantial changes to our process.
            3. Being tediously legalistic, this RFC currently triggers no tickets, so it can't be adopted.
            Show
            swinbank John Swinbank added a comment - Hi Josh — sorry, I didn't catch this one last week because I was travelling. I'd prefer that we walk this decision back for now for a few reasons: I'm concerned that references to version numbers exist in a number of baselined documents, and in the official project schedule. Changing the baseline can't be done without (at least) the approval of the DM-CCB and (possibly) the full project CCB. Per my comment on RFC-547 . Gabriele Comoretto and Leanne Guy are in the process of drafting a policy about release versioning and support. I'd like to give them chance to make a proposal before we adopt any substantial changes to our process. Being tediously legalistic, this RFC currently triggers no tickets, so it can't be adopted.
            Hide
            jhoblitt Joshua Hoblitt added a comment -

            OK – lets withdraw the RFC then.

            Show
            jhoblitt Joshua Hoblitt added a comment - OK – lets withdraw the RFC then.
            Hide
            swinbank John Swinbank added a comment -
            Show
            swinbank John Swinbank added a comment - Thanks Joshua Hoblitt !

              People

              Assignee:
              jhoblitt Joshua Hoblitt
              Reporter:
              jhoblitt Joshua Hoblitt
              Watchers:
              Frossie Economou, Gabriele Comoretto, John Swinbank, Jonathan Sick, Joshua Hoblitt, Krzysztof Findeisen
              Votes:
              0 Vote for this issue
              Watchers:
              6 Start watching this issue

                Dates

                Created:
                Updated:
                Resolved:
                Planned End: