Re: [v2] Overuse of response files?

From: Vladimir Prus (ghost_at_[hidden])
Date: 2005-05-25 09:42:11

David Abrahams wrote:

> Vladimir Prus <ghost_at_[hidden]> writes:
> 
>> On Tuesday 24 May 2005 20:12, David Abrahams wrote:
>>
>>> > And the most convenient way is core support, so that
>>> >
>>> > actions
>>> > {
>>> > cl.exe @@($(INCLUDES))
>>> > }
>>> >
>>> > will expand to
>>> >
>>> > cl.exe @some_file.rsp
>>> >
>>> > response file when command line length run of of some limit,
>>> > and expand to just
>>> >
>>> > cl.exe $(INCLUDES)
>>> >
>>> > otherwise.
>>>
>>> Yes, that would work.
>>
>> Do you think that's the best approach?
> 
> Well, I am reluctant to add more stuff to the core -- that'll just
> make porting to Python a bit harder -- but in order to get maximal
> usability we might need to do it. I think you have to watch out that
> your response files don't collide with some file the user
> intentionally generates with a .rsp extension, but I'm not sure how
> you'd do that. Anyway, that's not a new problem; it exists even for
> non-core approaches.
> 
> So, I'm on the fence about whether it's best. But it's at least as
> good as any other approach ;-)
> 
> Do you feel confident that "@@(...)" is a good syntax? 

Not yet. The original patch had something like "@(...)" as the syntax. The
entire construct was replaced by file name, and you could use any syntax
for response file one like. In that original patch, the first '@' in
@@(...) is not handled in any way and just copied verbatim. But if we want
to conditionally use response file, we need to replace entire @@(...) with
value in branches is the rsp file is not needed, and so we need to remove
the first '@'. This hardcoding of '@' is not nice, but I don't know better
approach.

> I presume you 
> could put anything at all in the parens, e.g.
> 
> @@(-I$(INCLUDES) -D$(DEFINES) $(>))

Yes.

>>> I would like to *not* use
>>>
>>> && del foo/bar/a.rsp
>>
>> Well, if response files are supported in core, there would be no need to
>> add any explicit 'del'.
> 
> Okay.
> 
>> If we change response files along the lines of my recent patch
>> (earlier in this thread), then.....
> 
> Sorry, I don't know what patch you're referring to. If you made it
> already, what's to change?

I think you've seen it already. Might be good enough already ;-)

- Volodya