Re: function-local variables in ltmain

[Ralf Wildenhues]

Hi Gary,

Welcome back!

* Gary V. Vaughan wrote on Tue, Aug 16, 2005 at 01:29:13PM CEST:
> Ralf Wildenhues wrote:
> >* Albert Chin wrote on Sat, Aug 13, 2005 at 09:56:43PM CEST:
> >>On Sat, Aug 13, 2005 at 10:40:01AM +0200, Ralf Wildenhues wrote:
> 
> >Just look at two examples: func_extract_archives uses variables named
> >my_*.  This is not safe: both its caller and any functions it may call
> >itself may overwrite these accidentally.  func_extract_an_archive OTOH
> >uses safe names f_ex_an_ar_*, but is ugly to the point of being
> >unreadable.
> 
> maintain an m4 map of function name to enumeration encoding, and name 
> the local variables localxx_var_name if output ugliness is a big issue. 

This breaks `eval'.  For the above-mentioned functions, that is no
issue.  But it prevents us from using any TAGVARs within such a function
with local variables, for example.

We might be able to cope with this limitation.  But it's a limitation
that
- "real" local variables do not have
- my solution does not have
- seems artifical to some point, and dangerous to another: in the m4sh
  source file, this might easily be overlooked!

> On the principle of looking at the .m4sh files way more often than the 
> generated libtool script, I don't think it is worth the trouble.
> 
> >I do _not_ want to eliminate use of globals.  All I want is an easier
> >method for ensuring this than regular grepping of the whole script.
> >I'm certainly not determined to use this particular idea.
> 
> I am :-)  Eventually.

No, please don't.  I have been sending small patches which change only
one thing at a time on purpose.  No need to over-engineer anything here.

> In my copious free time ;-)

Ahh, yes, ok, now I know how you meant that.  :->

Cheers,
Ralf