backing up my repositories, anything special needed?

[Giorgos Keramidas]

On 2007-08-13 10:00, Govind Salinas <govindsalinas at gmail.com> wrote:
> On 8/12/07, Giorgos Keramidas <keramida at ceid.upatras.gr> wrote:
> > I wonder how many people use Mercurial to get a faster, distributed
> > "overlay" on top of Perforce (or another centralized SCM) :-)
>
> Hi,
> I have been evaluating using a DSCM over Perforce myself for work,
> primarily so I could more easily collaborate with my co-workers (this
> of course implies getting them to use it too).  It really has come
> down to git vs hg.  I like hg because it is so much better supported
> on windows (git is making progress but it is still not complete
> support) and the lack of maintenance needed.  Unfortunately I am left
> with git that has good p4 support (can do go both ways) and hg which
> has good windows support.

Windows support is cool, and when TortoiseHg is stable enough for
"every-day users", I will definitely consider proposing a switch from
Perforce to Hg where I work :-)

> I have seen comments like these while browsing the hg archives and it
> seems like a lot of overhead to manually take the diffs from one place
> and then apply them to another.

I am not sure if what I do is the best way to handle this, but I wrote a
custom shell script which pulls changesets from ~/p4/project and pushes
them as Hg changesets to ~/hg/project/main with something like:

    p4-bringover ~/p4/project ~/hg/project/main //depot/project/main

The resulting changesets look like this:

    % hg log -l3 --template '{rev} | {author|email} {desc|firstline}\n'
    100 gkeramidas bringover 19210 from //depot/project/main
    100 gkeramidas bringover 19209 from //depot/project/main
    100 gkeramidas bringover 19208 from //depot/project/main
    %

Now that I see the p4import extension of Git I think it probably makes
more sense to avoid custom shell-based hackery like the script I've
written, and concentrate on writing an extension to "hg convert" :-)

> Does the list have any suggestions for me while I am still in the
> evaluation phase?

Not me, I'm afraid.  I'm not reallly a hardcore Python hacker who can
generate an extension like this very fast :-(