Malcolm Rowe wrote:
> On Mon, Jun 12, 2006 at 06:05:10PM +0100, Max Bowsher wrote:
>> Malcolm Rowe wrote:
>>> As far as I can see, the hack is no longer required on Cygwin (at least,
>>> the reversion doesn't seem to break anything), but I'm not aware how to
>>> reproduce the original problem, so I can't be sure.
>>> Could you confirm why you reverted the hack?
>> Because it now does it a better way.
>> Instead of exec-ing configure, in order to get the config.nice file
>> closed before configure tries to re-write it, now configure moves the
>> old config.nice aside and writes a new one.
>> This has the benefit that the path to configure doesn't get made
>> absolute, thus constraining the use of the config.nice file to the
>> particular current directory.
> Ah, thanks for explaining that - I hadn't noticed that it moved the
> old config.nice out the way first (and that also explains the failure
> I found - I used a config.nice generated by r19919 with a pre-19919 tree).
> Do you think we should add a comment to document the reason that we're
> moving the old config.nice out of the way first? It's not immediately
> obvious that we're doing it to appease Cygwin.
We could - I didn't because I figured that we were saving a previous
config.nice just because it was a generally nice thing to do, and it
just coincidentally helped Cygwin out.
> Also, should we add .config.nice.old to svn:ignore?
That, we should definitely do.
Received on Tue Jun 13 18:42:28 2006