On Sat, Jul 09, 2005 at 12:32:43PM -0400, Christopher Faylor wrote:
>On Sat, Jul 09, 2005 at 06:08:08AM -0600, Eric Blake wrote:
>>-----BEGIN PGP SIGNED MESSAGE-----
>>Hash: SHA1
>>
>>According to Christopher Faylor on 7/8/2005 11:02 PM:
>>> I don't get it. Certainly /cygdrive/c.exe doesn't exist so why would you
>>> try to link to it?
>>
>>The algorithm I used was this:
>>
>>1 - If source exists [determined by the return value of readlink
>>("source"), although just now I realized that for symbolic links, it
>>should really be the return value of readlink (concat (dirname ("dest"),
>>"source")) since symbolic links are relative to the destination, not the pwd]
>> 2a - if source is non-symlink and exists with a trailing . [determined by
>>access ("source.")] use "source", else use "source.exe", because one of
>>those two spellings must have made test 1 succeed
>> 2b - if source is symlink and exists with a trailing .lnk [determined by
>>readlink ("source.lnk")] use "source", else use "source.exe", because one
>>of those two spellings must have made test 1 succeed
>>
>>I think the cleanup needs to be along these lines - in steps 2a and 2b, do
>>a sanity check of also checking whether "source.exe" exists.
>
>I don't know why you'd ever just assume that .exe exists.
>
>Wouldn't it be safer to just do:
>
> struct stat st1, st2;
>
> stat ("foo", &st1);
>
> if (stat ("foo.exe", &st2) == 0 && st2.st_ino == st1.st_ino)
> return "foo.exe";
> else
> return "foo";
>
>You could add st_dev checking, too, if you were really paranoid.
You can also replace "stat" with "lstat" as appropriate, of course.
cgf
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
