[Omake] Re: [PATCH] Added a test case for .SUBDIRS problem

[Aleksey Nogin]

On 06.06.2007 03:01, David Kågedal wrote:

> David Kågedal <davidk at lysator.liu.se> writes:
> 
>> Here is another test case that broke my old omake scripts that used to
>> work a year ago or so.
>>
>> The problem is that the ".SUBDIRS: y" line is interpreted so that y is
>> assumed to be inside x for some reason.  It seems that the "current
>> directory" is getting exported by "export" or something like that.
> 
> After rereading the documentation, it seems that this is the
> documented behaviour of export.  But for .SUBDIRS sections, I don't
> see much sense in that.

I agree, seems like a bug - I've filed it as 
http://bugzilla.metaprl.org/show_bug.cgi?id=670

> By the way, is there a difference between
> 
>     .SUBDIRS: x
>         ...
> 
> and
> 
>     section
>         cd x
>         ...
> 
> ?

Yes. The .SUBDIRS form would:
  - Create a new set of phony "subtargets" inside x (as dependencies of 
the corresponding phony targets in the parent directory).
  - Do the same thing as "section" form does
  - Capture the environment that was created during the evaluation of 
the section.
  - Add the subdirectory "x" to the project, associating the captured 
environment with it. This will become the default environment for 
implicit rule lookup for targets in directory x

Aleksey

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