Re: Changes in update-game-score.c

[Eli Zaretskii]

> Date: Wed, 22 Jan 2014 19:32:01 -0800
> From: Paul Eggert <address@hidden>
> CC: address@hidden
> 
> Eli Zaretskii wrote:
> > Why do we suddenly care about it so much?
> 
> We got a bug report about it, which I fixed, and I noticed some other 
> bugs in the neighborhood, including undefined behavior and race 
> conditions which are dicey in a setuid program (it's bad PR at the very 
> least). A feature freeze shouldn't prevent us from fixing such bugs.

The bugfix should have fixed the bug that was reported, and that's it.
Even the first bugfix was already too much (you said it yourself:
fancy).  The rest is just superfluous and unneeded, as the program is
not important enough for us to waste any time on it.  (Why isn't it
implemented in Lisp, anyway?)

> The main culprit here was the w32 port's definition of fchmod as a noop 
> for Emacs, and its failure to define fchmod for non-Emacs executables. 
> We can't expect non-w32 experts to anticipate that sort of tricky 
> situation flawlessly every time.

No one requested you to be the expert, you are missing the point.

You never explained why the switch from chmod to fchmod was needed
anyway.  Do we really care about race conditions here?  What, the same
user will be updating her score simultaneously from 2 sessions?  Come
on!

Feature freeze means restraint: do not try to "fix" that which has
been happily "broken" for ages.