Re: amber-developers: gnu make was [sbrozell.scripps.edu: Re: AMBER: Installation amber 9 on IBM SP4]

From: Scott Brozell <sbrozell.scripps.edu>
Date: Sat, 26 Aug 2006 14:40:06 -0700

Hi,

On Sun, 27 Aug 2006, Xuebin Qiao wrote:

> Maybe, it's wise to seperate the computational logic as well as building
> processes of GUI-dependant and GUI independant codes at certain stage, just
> like the branching of antechamber. If gleap is really based gtk+opengl (I
> don't know exactly, cuz' I didn't see the codes) or something alike, its
> portability and scalebility wouldn't be as good as other GUI-independant
> computational codes.
>
On Fri, 25 Aug 2006, David A. Case wrote:

> Scott Brozell wrote:
>
> > It looks like $< is not being evaluated correctly.
>
> I'm really tempted to just say that that Amber requires GNU make. This would
> allow us to do so many things much easier (especially conditional execution),
> plus would eliminate the royal pain of trying to keep things compatible with
> three-decade old UNIX make programs that we rarely test.

If you distribute gnu make and if it could be merged into the build process
as simply as changing the path of the make, eg:

setenv AMBERHOME
./configure bla
$AMBERHOME/exe/gnumake bla

then this might solve some annoying build issues.

Regardless, we should be doing more testing as this eliminates only one
small set of build problems; there are many other sets, such as compiler
issues, parallel issues, etc. Those should be tested for on IBM and a
couple of other platforms that are picky and not as commonly used by
amber developers.

Eliminating all compiler warnings would enable us to crank up the warnings
for daily buildings; that would make it easier to spot potential problems
at the time of occurence.

> Note that gleap is really going to require a modern build environment.

Why ?

> But I'm open to counter-arguments...we do have to put up with IBM since they
> make the machines people want to work on.
>
> [Note: this is only for Amber10; we still have to patch Amber 9].

Ok, I'll post a bugfix.

> ... there have been two really clean,
> consistent models of programming so far:
> the C model and the Lisp model.
> These two seem points of high ground,
> with swampy lowlands between them.
> --Paul Graham

I would like to know the context of this quote.
As it stands I disagree with this quote in many ways.

Scott
Received on Sun Aug 27 2006 - 06:07:32 PDT
Custom Search