[PATCH] Harden uses of os.system

[Matt Mackall]

On Sat, Jun 25, 2005 at 04:10:40PM -0700, Bryan O'Sullivan wrote:
> On Sat, 2005-06-25 at 12:22 -0700, Matt Mackall wrote:
> 
> > Any particular reason we're printing the warning down here rather than
> > propagating it up? I'd prefer the latter.
> 
> As in, throw it as a parameter to the Abort, and print it at the top?
> Sure, that's fine.

Yes.
 
> > These bits snuck in.
> 
> Bugger.  That should have been in the "init dest directory" patch.  I'm
> not too clear on which files do and don't make it into a commit, yet.

a) everything listed after hg commit

or

b) everything not listed as ? in 'hg status'

> A certain other distributed revision control system has a useful GUI
> tool that tells you exactly what is going to be in a commit, and shows
> you the diffs, prompting you for the commit comments.  If I'd had
> something like that, I wouldn't have made this mistake.

Are you using hgeditor? It basically does that, but without the GUI
bits. Though TAH broke it so it now shows diffs of everything rather
than just what's in the commit.

> I can cook such stuff up, nice and fast.  Will you take patches that add
> GUI components?  They'll depend on PyQt, since it's got a sane API, is
> portable between X, Macs, and Windows, and is present on most Linux
> boxes.

As it happens, I was just tinkering with a GUI this morning. But mine
uses GTK+ and GLADE. Sigh.

-- 
Mathematics is the supreme nostalgia of our time.