New run version with patches for Windows 7

[Charles Wilson]

[consolidated reply]

Corinna Vinschen wrote:
> Oh, right.  It won't work quite correctly under 1.5 either.  Why do
> we need a 1.5 aware version again?  Anyway, if you look into the
> patch, you'll see that it is very simple to keep the old code in
> conditionally.  What about this one?  It only leaves Cygwin 1.5 behind
> which is our intention anyway.

Now this, I like.  I'll probably adapt something along these lines for
run2.  As for the original 'run', I'll leave that decision to the actual
maintainer.

Corinna Vinschen wrote:
> Why do we need a mingw version?

"We" don't.  But "run" is not only a cygwin project -- just like
coreutils or bzip2 is not solely focused on cygwin.  I know that Colin
Harrison ships a version of run with his XMing distribution, and I've
seen it in various other places around the net, as well -- compiled as a
native program.  I think I even remember seeing one version compiled
using MSVC.

Now, AFAIK, Alexander wears (wore?) two hats:
  1) "upstream" maintainer of the run package, supporting various (okay,
two -- maybe three) different platforms: cygwin, mingw, and maybe native
win32 msvc.
  2) "cygwin" maintainer who provides the official cygwin package for
the run utility.

"We" as in #2 don't need mingw support [*].  However, whether Alexander
continues to feel the need to support mingw while wearing hat #1, I
don't know.  That's up to him.

[*] back when I maintained run, there were a few releases in which I
experimented with compiling it using -mno-cygwin, even for the cygwin
release.  It didn't work out well.

Corinna Vinschen wrote:
> Christopher Faylor wrote: 
>> Since we don't have a package maintainer for run, why not just
>> make this change and release it for 1.7?
> 
> We don't know yet if we don't have a package maintainer.  I'd like
> to wait if Alexander replies in the next few days.

If not, I'll take it back. As part of developing run2
(run_gpl.c/run_gpl.h) I've had quite a close look at the code, and have
a few ideas for refactoring.  I'd give Alexander a few days to respond, tho.

>> Or, if Chuck's run2 is superior, why not just call it "run" for
>> 1.7 and avoid releasing this version at all?
> 
> run2 is no drop-in replacement for run since it requires XML files
> for the configuration.

Exactly. More capabilities comes at a price (it may be just my
imagination, but I think run2 might also be just a smidgen slower; it
does have to validate and parse that xml, after all).  I think there's
room -- and need -- for both versions within the distribution.

--
Chuck