Wrong BugID (whoops!)

Dmitry Samersoff dmitry.samersoff at oracle.com
Tue Jul 9 14:37:51 PDT 2013


On 2013-07-10 00:56, mark.reinhold at oracle.com wrote:
> 2013/7/9 5:50 -0700, dmitry.samersoff at oracle.com:
>> 1.
>> For cases like this I would like to have a file counting problematic
>> changesets e.g. ProblemChangesetList.txt, otherwise it's hard to
>> restore the history after couple of years.

Sorry! "Problematic" is not the best word here.

> I don't understand.  The whole point of removing problematic changesets
> is so that they're never part of the long-term history.  Why do you need
> a list of them?
> In any case, if you do need a list then the hashes of all blacklisted
> changesets in OpenJDK repositories are in jcheck.py itself.  (There's
> a separate server-side list for Oracle-internal repositories.)

As far as I understand, we don't plan to remove changeset - just
manually close bug in JBS.

In this case we keep a wrong bug id in history so attempt to understand
what caused the changes couple of years later become an adventure.

I would like to have something not very formal that looks like:

(wrong bug id)
4896:97c5acae48be s/7007040/7007044/

(hotspot and corresponding jdk changes has different bug id)
4896:97c5acae48be see also jdk 7007044



>> 2.
>> Starting from Mercurial 2.2
>> hg has --amend command that allows to perform limited but safe editing
>> of the history. Never tried it so not sure it helps in this case.
> It doesn't help in this case.  That option only works in a local
> repository, on changesets that haven't yet been pushed anywhere else.
> - Mark

Dmitry Samersoff
Oracle Java development team, Saint Petersburg, Russia
* I would love to change the world, but they won't give me the sources.

More information about the hotspot-dev mailing list