Re: Performance issues with Cons

[email protected] Mon, 7 Oct 2002 10:48:59 -0700
Newsgroups gmane.comp.gnu.cons.general
Message-ID <[email protected]>
Warren wrote:
I certainly understand that ---  however, as I said in my other post, I'd 
like a
way for developers to say: "I'm pretty sure that the dependencies haven't
changed since the last time you built them - please don't rescan 
everything".

Here's my understanding, hopefully someone will correct me if it's wrong. 
In theory this is called pruning. The cons docs say you can use  a regex 
to "prune" what cons will build. I have tried it without much success, but 
that could definitely be pilot error. However, we have implemented a 
different strategy. Our strategy is to only let cons know about what we 
want to build. So it is not doing full dependency checking. So instead of 
directing cons to filter what it knows about, we are filtering what is 
input to cons.

Does that help?

David Weidenkopf




[email protected]
Sent by: [email protected]
10/07/2002 10:21 AM

 
        To:     David Weidenkopf/ATL-BTL/MS/PHILIPS@AMEC
        cc:     [email protected]
        Subject:        Re: Performance issues with Cons
        Classification: 






> The performance penalty is the cost of correctness. Cons is analyzing 
every
dependency before it does anything.

I certainly understand that ---  however, as I said in my other post, I'd 
like a
way for developers to say: "I'm pretty sure that the dependancies haven't
changed since the last time you built them - please don't rescan 
everything".

>You might find the scripting capabilities of gnu-make lacking. We looked 
at
using gnu-make, but we got our Cons
> prototype working so quickly we abandoned the thought.

I've used gnu-make enough to know that I really don't like it... Switching 
to
gnu-make is really quite low on my list of priorities...  If we can't 
figure out
how to do this with cons, my inclination is to suggest we switch to scons, 
which
already supports a feature like this...

Frankly, I'm tempted to suggest switching to scons anyways, since it 
doesn't
look like there's a lot of interest in working on cons, but scons still 
seems to
have a fairly active development community...

Warren




_______________________________________________
[email protected]
http://mail.gnu.org/mailman/listinfo/cons-discuss
Cons URL: http://www.dsmit.com/cons/