Re: Moving pkgsrc-wip away from SourceForge

[Andrew Cagney <andrew.cagney%gmail.com@localhost>]

On 14 July 2015 at 19:05, John Nemeth <jnemeth%cue.bc.ca@localhost> wrote:
>
>      So if somebody makes an update to one of the several versions
> of PostgreSQL while I'm making an update to one of the several
> versions of MySQL, my push should fail?!?  That's a totally ridiculous
> notion.

If the two projects are truly independent then why put them in the
same repository? (the obvious response is convenience)

More seriously, when I type:
   svn commit
the operation atomic - I'm simply throwing dice (ever reliable
wikipedia points out this is atomic in the database transaction
sense).  I've no assurance that the result is valid (I suspect by now
there is a flag to change this).  To me, that is an equally ridiculous
notion :-)  This all likely doesn't matter for pkgsrc-wip, but it does
(I argue) when things move to pkgsrc and tree-wide changes are being
made.

Taking a step back, though, which of the criticisms of CVS
(https://en.wikipedia.org/wiki/Concurrent_Versions_System#Criticism)
are a real cause for concern, and which does SVN address (Some
versions of CVS tag commits meaning that it is possible to uniquely
identify them)

Perhaps the only change needed is for the repository to move.

Andrew

PS: for git (HG equiv?), standard operating procedure when there are
small unrelated changes is: "git fetch origin ; git rebase origin ;
git push]" (or just "git pull --rebase").  When there's a conflict or
race you're in clear control over how things get resolved.