Re: I think binary-branch is almost ready for play time

[Bruce Korb]

Alexandre Oliva wrote:
> > Run this command:
> 
> >    autogen -L ${prefix}/share/autogen --save=$HOME/.autogenrc
> 
> > where "${prefix}" is whereever you installed it.
> 
> Ok, that got me a little bit further.

Good.  I'll add another search directory for the next rev:

  $$/../share/autogen

> Now I get:
> 
> autogen -T ltstr.tpl ltstr.def
> ERROR: Unbound variable: string->c-name!
> AutoGen ABENDED in template ltstr.tpl line 10

Ah, now *that* means you need to upgrade.
I must have added that function after 5.0.9.
I think you could also emulate with:

  (define string->c-name! (lambda (x)
     (set! x (shellf "echo '%s'|sed 's/[^a-zA-Z0-9 \t]/_/g'" x)) ))

or some such.  Probably easier to upgrade.  :-)

> Then, as soon as you're satisfied that the binary-branch is in a
> stable state, tag it too, using some tag name that indicates it can
> generate an ltmain.in identical to that of MLB.  Then, if anyone
> inadvertently goes and installs a patch in MLB, you can easily tell by
> diffing the branch with the branchpoint, and if someone makes a change
> in the binary branch, you can easily tell by comparing with the tag
> you'll create in this branch.  How's that?

Truely yummy.  ;-)

> > So, step 2 is actually a process.
> 
> I understand.  I just want to see a first example of where step 2 can
> take us to be convinced it's the way to go.  I still have trouble
> understanding how to achieve that with AutoGen.

Make you a deal.  If you do the CVS hocus-pocus, I'll pound out
an example with, say, the "execute" code, ok?  :-)  Thanks!