[darcs-users] darcs patch: make reading the pending lazy in summary mode

[David Roundy]

On Fri, Apr 25, 2008 at 08:09:59AM -0700, Jason Dagit wrote:
> Warning:  I'm not 100% sure that this should be applied, see below.
> 
> This might be a fix for Issue814.  It skips a read of pending when
> in summary mode.  Since not all reads of pending are skipped I don't
> think the problem is that reading pending is too strict, but there
> is definitely some unwanted strictness here.  I suspect it's a
> problem with what we do after getting the pending patches.
> 
> All tests pass, but I'm worried that in some corner cases of
> 'darcs whats -s' that pending changes could be reported
> incorrectly.  Maybe someone could look this over and turn it
> into a real fix for Issue814?
> 
> Thanks,
> Jason
> 
> Fri Apr 25 00:49:07 PDT 2008  Jason Dagit <dagit at codersbase.com>
>   * make reading the pending lazy in summary mode

This looks like a wrong change to me.  It may not currently introduce a
bug, but your concerns are correct.  At a minimum it introduces a latent
bug that could bite our users if we ever add the --summary flag to any
command that affects our repository.  It also changes the behavior of darcs
whatsnew to be less accurate.

>  slurp_pending :: RepoPatch p => Repository p -> IO Slurpy
> -slurp_pending repo = do
> +slurp_pending repo@(Repo _ opts _ _) = do
>    cur <- slurp_recorded repo
> hunk ./src/Darcs/Repository/Internal.lhs 234
> -  pend <- read_pending repo
> +  pend <- if Summary `elem` opts
> +            then return ( NilFL :: FL Prim)
> +            else read_pending repo
>    case apply_to_slurpy pend cur of
>      Just pendcur -> return pendcur

If we ever add the --summary flag to (for instance) record, push or pull,
this will corrupt repositories.  I could easily imagine wanting to see a
summary of each patch that I'm prompted to pull, as if I had hit 'x'.  So
no, this is definitely not the way to go.

The real fix for Issue814 is to ignore any changes to newly added files
when doing whatsnew -s, and this change needs to happen in diff.  Gwern's
idea of making is_funky faster is always good (since it speeds up many
darcs commands, if only a little), but I don't think it touches the real
problem, which is that we shouldn't be opening these files at all.
-- 
David Roundy
Department of Physics
Oregon State University