[Ecls-list] Conflicting init_ functions when building libraries

Goffioul Michael goffioul at imec.be
Thu Mar 16 23:59:07 UTC 2006

> Thanks for your suggestion. I tried that and it partly solved 
> my problem, but only in exchange for another one.
> When i compile the system, i produce fasl and object files in 
> parallel, which i have to do with two compile-file calls, 
> because i can not inhibit the deletion of the intermediate 
> object file (why?).
> I then load the fasl files immediately to have a lisp image 
> that reflects the state of the already compiled/loaded files. 
> The dynamic lib that is produced this way does exactly the 
> right thing, namely what the interpreter would do when all 
> the fasl files were loaded in the right order.
> Setting si::*init-function-prefix* forces me to produce only 
> the object files since this variable does not seem to be 
> considered when building fasls, which results in an error 
> when loading them. But the library that results from this 
> computation behaves differently (wrongly) from the one that 
> subsequently loads the fasls.
> Any suggestion?

I'm not sure I did follow everything...

During the compilation process, isn't is possible for you to load the
source LISP code instead of a FASL file? This is something I used to
compile our own framework: each source file is compiled into an object
(using :system-p set to true) and then the source file is loaded.


More information about the ecl-devel mailing list