#1942 Mercurial file browser shows wrong thing NEEDS SCRIPT

v1.0.0
closed
Rick Copeland
sf-4 (350)
General
nobody
2015-08-20
2011-04-13
Rick Copeland
No

Example: https://sourceforge.net/p/forge-migrate4/code-0/ci/5f4c74506c370494203e0371431a34d5b1ee4b3c/tree/

This should actually have a number of directories, but all it has is a single file.

Discussion

  • Dave Brondsema

    Dave Brondsema - 2011-04-14
    • size: --> 4
     

  • Dave Brondsema

    Dave Brondsema - 2011-04-18
    • status: open --> in-progress
    • assigned_to: Dave Brondsema
     

  • Dave Brondsema

    Dave Brondsema - 2011-04-18

    I'm not sure what's going on. If I clone the repo within production Allura, the problem persists: https://sourceforge.net/u/brondsem/test-hg/

    But if I "hg clone" from forge-migrate4 and push it into a sandbox repo, it all is fine. Also, cloning directly from the original classic repo like the migration script does worked fine:

    h.set_context('test4')
c.project.install_app('Hg', mount_point='fork-prod', mount_label='fork-prod', init_from_url='http://forge-migrate4.hg.sourceforge.net:8000/hgroot/forge-migrate4/forge-migrate4')
     

  • Dave Brondsema

    Dave Brondsema - 2011-04-18

    The commits' file diffs, and tree objects differ between local and production. They seem more correct locally, but still not all correct. Commit tip is:

    db.repo_object.find({object_id: '5f4c74506c370494203e0371431a34d5b1ee4b3c'})

     

  • Dave Brondsema

    Dave Brondsema - 2011-04-18

    I can't find where Commit.tree is set for Hg. refresh_commit's fake_tree seems plausible, but doesn't appear to run when importing a new tree

     

  • Dave Brondsema

    Dave Brondsema - 2011-04-18
    • status: in-progress --> blocked
     

  • Dave Brondsema

    Dave Brondsema - 2011-04-25
    • status: blocked --> in-progress
    • milestone: may-05 --> apr-28
     

  • Dave Brondsema

    Dave Brondsema - 2011-04-29 
    :::text
> use pyforge
> apr4id = new ObjectId( Math.floor( new Date(2011,4-1,4).getTime() / 1000 ).toString(16) + "0000000000000000" ) 
> db.repo_object.remove({_id: {$gt: apr4id}}) 

# specific projects:
allurapaste script production.ini ../scripts/refresh-all-repos.py -- necessitas/code
# all:
allurapaste script production.ini ../scripts/refresh-all-repos.py
     

  • Dave Brondsema

    Dave Brondsema - 2011-04-29

    forge:db/1942

    • new tests cover the broken functionality; hard to duplicate manually
    • to fix the data in production, we'll run the above commands. Apr 1 is used because the regression happened then, with commit f461c6 which removed __repr__
    • you can test the above commands by importing a repo of any type, then deleting some of the repo_objects (e.g. above some _id) and then running the refresh command.
     

  • Dave Brondsema

    Dave Brondsema - 2011-04-29
    • summary: Mercurial file browser shows wrong thing --> Mercurial file browser shows wrong thing NEEDS SCRIPT
    • status: in-progress --> code-review
    • assigned_to: Dave Brondsema --> Rick Copéland
     

  • Rick Copeland - 2011-05-03
    • status: code-review --> validation
     

  • Dave Brondsema

    Dave Brondsema - 2011-05-09
    • status: validation --> closed
     

Log in to post a comment.