Re: efficient use of auto-builder machines (was Re: Bug#43787: changed title, and remade the proposed change)
On Wed, Sep 08, 1999 at 07:07:15AM -0400, Raul Miller wrote:
> On Tue, Sep 07, 1999 at 09:37:19PM -0400, Ben Collins wrote:
> > Umm, how do you see your hack as a speed gain when it requires every
> > invocation of gcc to also invoke perl?!
> I guess that means you didn't read the rest of the message.
> It's trivial to rewrite in C, and I offered to do so.
It's also trivial to recompile the gcc binary so it does not pass -g to
any of the compilers such as cc1 and cc1plus, but this is still a gross
hack that completey changes the way the programs are _supposed_ to be
built by the maintainer. It's a serious hack, and IMO, would be really
dangerous, and I definetly would not implement any such thing just for the
sake of not screwing things up for an entire architecture.
> > So you are saying that my proposal, one which helps define some build specs,
> > is a hack, and your suggestion to write a wrapper around gcc to change the
> > way packages are built by default is not?
> No, it's very definitely a hack.
> However, like I explained before, it's a hack that doesn't define a new
> hack interface which would have to be supported when we do things the
> right way.
Your opinion is not backed by any solid facts. 1) You don't run an
autobuilder, so please don't start making suggestions on something you
have no idea about, 2) You already said you don't like it because you
don't see any benefit, that's not my problem, I don't care if you don't
get any benefit from my proposal. Find something else to do with your
time, other than trying to stop would-be-progress.
> > This is where I start to ignore everything you say.
> Apparently you started to ignore what I had to say at least a couple
> messages back?
Yeah, maybe I did. It's hard to stay focused on unfocused rants.