Loading mysterium

Loading mysterium

[Stefan Israelsson Tampe]

Hi,

Consider macro m1 in module a) is a macro writer macro
In module b) m1 is used to create m2

Now loading module b) and using fails in some circumstances.

What I do now works is to touch a) and b) and then import b)
causing them to be recompiled. Then using b) is fine.

Any clues?

p.s. savannah seams to down, cannot check in or check dup this issue.

/Stefan

Re: Loading mysterium

[Ludovic Courtès]

Hi Stefan,

Stefan Israelsson Tampe <stefan.itampe@gmail.com> writes:

> Consider macro m1 in module a) is a macro writer macro
> In module b) m1 is used to create m2
>
> Now loading module b) and using fails in some circumstances.

Can you provide more details (error messages, etc.), and perhaps an
example to reproduce the problem?

BTW, I think the Right Term is “macro-producing macro”.  ;-)

Ludo’.

Re: Loading mysterium

[Stefan Israelsson Tampe]

Thanks for your reply Ludo,

Let me try to debu a little and climbe above the 
noise level. I do have some problems I did not pay attention
to before. I must clean that first.

/Stefan
On Tuesday, December 21, 2010 06:52:23 pm Ludovic Courtès wrote:
> Hi Stefan,
> 
> Stefan Israelsson Tampe <stefan.itampe@gmail.com> writes:
> > Consider macro m1 in module a) is a macro writer macro
> > In module b) m1 is used to create m2
> > 
> > Now loading module b) and using fails in some circumstances.
> 
> Can you provide more details (error messages, etc.), and perhaps an
> example to reproduce the problem?
> 
> BTW, I think the Right Term is “macro-producing macro”.  ;-)
> 
> Ludo’.

Re: Loading mysterium

[Andy Wingo]

On Sat 18 Dec 2010 10:45, Stefan Israelsson Tampe <stefan.itampe@gmail.com> writes:

> Consider macro m1 in module a) is a macro writer macro
> In module b) m1 is used to create m2
>
> Now loading module b) and using fails in some circumstances.

...the circumstance being, the interface that m2 was expanded against
changed.  In theory this should render the compiled module b) too-old,
with respect to its dependencies -- which is to say, not just b.scm, but
also a.scm (or more generally, the module a).

Guile doesn't handle this properly.  It assumes that if b.go is newer
than b.scm, b.go is usable.  That is not the case, generally.

A proper solution would have b.go encode the modules that it depends
on, and check against those modules.

Andy
-- 
http://wingolog.org/