[darcs-users] New argument for symlink support

Eric Kow kowey at darcs.net
Wed Dec 2 10:45:47 UTC 2009


Hi Henrik,

On Tue, Dec 01, 2009 at 22:09:01 +0100, Henrik Hjelte wrote:
> At least I have not seen it mentioned at the deferred ticket
> "wishlist: manage symlinks under version control".
> http://bugs.darcs.net/issue820
> 
> The absence of symlink-support makes it impossible to migrate a lot of
> repositories to darcs from other revision control systems. Adding a
> script that manages symlinks (like update-symlinks.sh) script is not a
> solution for that.

I sympathise with your frustration on this because I also use many
symlinks in projects I want to version control and I also have a user
who is reluctant to migrate an old project from no-version-control to
darcs because of the lack of symlink support.

In the absence of symlink support, I suggest having a look at the
FAQ for some workarounds, which while far from satisfactory, may
be better than nothing:
  http://wiki.darcs.net/FrequentlyAskedQuestions#can-darcs-handle-symlinks

As Jason points out, it would be useful if you could put some more
thought into how symlink support would work concretely, say on
  http://wiki.darcs.net/Ideas/Symlinks

And as Petr says, this is not work that we can do lightly.  We have to
be very careful about how we do this sort of work because we are
concerned not only about backward compatibility (in this case, making
sure that things either work with old darcs or more likely, just fail
gracefully), but also making sure that we're not stuck with a legacy of
supporting ill-designed mechanisms (consider our preferences system).

> Personally I think this is the weakest spot of darcs. A critical
> feature. It can not be an uncommon wish to standardize on one version
> controlling system only, then you need to be able to migrate from
> others to darcs.

So while I understand your arguments for why this is a critical feature,
I stand by my deferral of the ticket above.  It does not mean that we
are not going to work on this; it just means that we must first build up
up the resources (time, attention, hackers, goodwill) to plunge into
this long term work.  So please bear with us as we focus our attention
elsewhere (*)!

Thank-you very much for arguing that this is a critical feature.  I
think it may be fair to add this to a list of Darcs 3 must haves
  http://wiki.darcs.net/Darcs3Wishlist
Max may also want to think about what goes on there.

Best regards,

Eric

(*) My current medium-term (1 or 2 years) objective, one which I think
    other Darcs hackers may share, is a Darcs which is "fast enough".
    We certainly do not need to be the fastest revision control system
    on the block, nor optimise down to the last microsecond, but we must
    be fast enough to avoid the most common causes of frustration, and
    to change the popular Darcs Story (something about pie in the sky
    "scientists" with their impractical theories and real world
    "engineers" who make things work).  Patch theory does not
    necessarily condemn us to slow revision control; we just need to
    take the time out to do the practical work.

-- 
Eric Kow <http://www.nltg.brighton.ac.uk/home/Eric.Kow>
PGP Key ID: 08AC04F9
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: Digital signature
URL: <http://lists.osuosl.org/pipermail/darcs-users/attachments/20091202/a8c03177/attachment.pgp>


More information about the darcs-users mailing list