apr-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Brian Havard" <bri...@kheldar.apana.org.au>
Subject Re: cvs commit: apr/build find_apr.m4
Date Sun, 10 Feb 2002 14:49:39 GMT
On Fri, 8 Feb 2002 18:16:57 -0800, Greg Stein wrote:

>On Fri, Feb 08, 2002 at 10:46:46PM +1000, Brian Havard wrote:
>>...
>> >  -    elif test -f "$withval/apr-config"; then
>> >  +    elif test -x "$withval/apr-config"; then
>> >         apr_found="yes"
>> >         apr_config="$withval/apr-config"
>> >       elif test -x "$withval" && $withval --help > /dev/null 2>&1
; then
>> 
>> Was there any actual need to do this? I ask because it breaks the OS/2
>> build due to the fact that OS/2 has no 'x' bit in the file system. "test
>> -x" is only true if the file has a .exe extension which apr-config
>> obviously does not.
>
>Well, the apr-config file is generated, so it normally comes out without an
>executable flag. We then have a chmod +x (see at the end of configure.in)
>which should get run during the generation step.
>
>I switched to -x to avoid the case of an apr-config that wasn't make
>executable, and to avoid thinking we have a valid apr-config and then try to
>run the thing.
>
>That said: I'd be fine with loosening it up to just -f, as long as you put
>in some "dnl" comments on why we chose -f rather than -x. (otherwise, three
>years from now, somebody will go and patch it to put the -x flags back in
>there :-)

Either that or make the test switch adjust appropriately for the platform.
I'll see if I can figure out a nice way to do that.

-- 
 ______________________________________________________________________________
 |  Brian Havard                 |  "He is not the messiah!                   |
 |  brianh@kheldar.apana.org.au  |  He's a very naughty boy!" - Life of Brian |
 ------------------------------------------------------------------------------


Mime
View raw message