Return to bulktracker.mdwn CVS log

Up to [NetBSD Developer Wiki] / wikisrc / projects / project

Describe the bulk build tracker that's been talked about on and off
for the last few years. It's too bad I didn't think of putting this
here before the gsoc deadline...

    1: [[!template id=project
    2: 
    3: title="Bulk Build Tracker Application"
    4: 
    5: contact="""
    6: [tech-pkg](mailto:tech-pkg@NetBSD.org)
    7: """
    8: 
    9: category="pkgsrc"
   10: difficulty="medium"
   11: 
   12: description="""
   13: Currently, bulk build results are sent to the pkgsrc-bulk mailing
   14: list. To figure out if a package is or has been building successfully,
   15: or when it broke, one must wade through the list archives and search
   16: the report e-mails by hand. Furthermore, to figure out what commits if
   17: any were correlated with a breakage, one must wade through the
   18: pkgsrc-changes archive and cross-reference manually.
   19: 
   20: The project is to produce a web/database application that can be run
   21: from the pkgsrc releng website on NetBSD.org that tracks bulk build
   22: successes and failures and provides search and crossreferencing
   23: facilities.
   24: 
   25: The application should subscribe to the pkgsrc-bulk and pkgsrc-changes
   26: mailing lists and ingest the data it finds into a SQL database. It
   27: should track commits to each package (and possibly infrastructure
   28: changes that might affect all packages) on both HEAD and the current
   29: stable branch, and also all successful and failed build reports on a
   30: per-platform (OS and/or machine type) basis.
   31: 
   32: The web part of the application should be able to retrieve summaries
   33: of currently broken packages, in general or for a specific platform
   34: and/or specific branch. It should also be able to generate a more
   35: detailed report about a single package, containing for example which
   36: platforms it has been built on recently and whether it succeeded or
   37: not; also, if it is broken, how long it has been broken, and the
   38: history of package commits and version bumps vs. build results. There
   39: will likely be other searches/reports wanted as well.
   40: 
   41: The application should also have an interface for people who do
   42: partial or individual-package check builds; that is, it should be able
   43: to generate a list of packages that have not been built since they
   44: were last committed, on a given platform or possibly on a per-user
   45: basis, and accept results from attempting to build these or subsets of
   46: these packages. It is not entirely clear what this interface should be
   47: (and e.g. whether it should be command-line-based, web-based, or what,
   48: and whether it should be limited to developers) and it's reasonable to
   49: expect that some refinements or rearrangements to it will be needed
   50: after the initial deployment.
   51: 
   52: The application should also be able to record cross-references to the
   53: bug database. To begin with at least it's reasonable for this to be
   54: handled manually.
   55: 
   56: This project should be a routine web/database application; there is
   57: nothing particularly unusual about it from that standpoint. The part
   58: that becomes somewhat less trivial is making all the data flows work:
   59: for example, it is probably necessary to coordinate an improvement in
   60: the way bulk build results are tagged by platform. It is also
   61: necessary to avoid importing the reports that appear occasionally on
   62: pkgsrc-bulk from misconfigured pbulk installs.
   63: 
   64: Note also that "OS" and "machine type" are not the only variables that
   65: can affect build outcome. There are also multiple compilers on some
   66: platforms, for which the results should be tracked separately, plus
   67: other factors such as non-default installation paths. Part of the
   68: planning phase for this project should be to identify all the
   69: variables of this type that should be tracked.
   70: 
   71: Also remember that what constitutes a "package" is somewhat slippery
   72: as well. The pkgsrc directory for a package is not a unique key;
   73: multiversion packages, such as Python and Ruby extensions, generate
   74: multiple results from a single package directory. There are also a few
   75: packages where for whatever reason the package name does not match the
   76: pkgsrc directory. The best key seems to be the pkgsrc directory paired
   77: with the package-name-without-version.
   78: """
   79: ]]