[Omake] sections & implicit rules

[Aleksey Nogin]

On 26.03.2007 10:46, Aleksey Nogin wrote:

> On 25.03.2007 14:13, Jonathan Roewen wrote:
> 
>> Ohh, could an explicit dependency in a parent OMakefile cause the 
>> problems?
>>
>> In the parent OMakefile, I have:
>>
>> stdlib/pervasives.cmi: VFS.cmi stdio.cmi
>>
>> This occurs after the .SUBDIRS directive, so I thought it would be
>> okay...
> 
> Wow, it turns out that such a line would indeed affect the outcome! This 
> is not what I would expected at all - must be a bug! Thanks a lot for 
> finding this - I've filed it as 
> http://bugzilla.metaprl.org/show_bug.cgi?id=656

What seems to be happening is:
  1) When the target is pinned several times, the _last_ one wins 
(regardless of where it is).
  2) If the winning one is outside the target's directory, the default 
environment for the target's directory is used.

The right thing to do would be to simply ignore the dependency specs 
given outside the target's directory, for the purposes of finding the 
variable environment for the target. In addition, we should probably 
give a warning if several different candidate environments exist for a 
single target...

Aleksey

-- 
Aleksey Nogin, Research Staff Member
Advanced Technologies Department, Information & System Sciences Lab
HRL Laboratories, LLC, Malibu, CA