Re: ASDF: module and system level dependencies don't work right?

Hi Todd,

I like things the way they are though documenting the behavior  
explicitly would be good. To my mind each module is a separate  
"piece" and the flow of dependencies should stop at each boundary. On  
the other hand, dependency ought to flow between the files within a  
module. I don't have a good justification for this but it seems  
reasonable enough to me and fits my mental model.


On Dec 7, 2005, at 11:40 AM, Todd Sabin wrote:

>
> The :depends-on attribute seems to work differently with modules and
> systems than it does with individual source files.  If you have a  
> system
> like this:
>
> (asdf:defsystem system1
>   :components
>   ((:file "package")
>    (:file "foo" :depends-on ("package"))
>    (:file "bar" :depends-on ("package" "foo")))
>   :depends-on ())
>
> then modifying package.lisp and doing a load-op will cause all three
> files to be compiled and loaded.
>
> However, if you have a system like this:
>
> (asdf:defsystem system2
>   :components
>   ((:module package
>             :components ((:file "package")))
>    (:module foo
>             :components ((:file "foo"))
>             :depends-on (package))
>    (:module bar
>             :components ((:file "bar"))
>             :depends-on (package foo)))
>   :depends-on ())
>
> then modifying package.lisp and doing a load-op will cause only
> package.lisp to be compiled and loaded.  (And similarly for system
> level dependencies.)
>
> Another way to say it would be that for modules and systems,
> :depends-on seems to operate more like a ":requires" operator.
>
> Is this intentional?  I've assumed it's a bug, and have "fixed" it
> locally, but my "fix" ended up changing a lot of stuff, and almost
> certainly broke some undocumented features.  (I couldn't figure out
> how the :force option was supposed to work, for example.)  I'm pretty
> sure it also ends up doing more loading than necessary now, but I
> prefer too much loading to not enough compiling.
>
> If someone can confirm this is a bug, and tell me how things like
> :force are supposed to work, I can try to clean up my patch.  I also
> created a few more complicated system definitions for testing.  I
> suppose it'd be nice to add them to the test suite, but I haven't done
> that yet...
>
> Thanks,
>
>
> Todd
>
> p.s.  If anyone wants my half-baked patch as is, let me know.  It'd be
> against an asdf.lisp from a few revisions back of cvs head, but I've
> confirmed the behavior above is still present in head.
>
> -- 
> Todd Sabin                                           
> <[email protected]>
>
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc. Do you grep through  
> log files
> for problems?  Stop!  Download the new AJAX search engine that makes
> searching your log files as easy as surfing the  web.  DOWNLOAD  
> SPLUNK!
> http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click
> _______________________________________________
> cclan-list mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/cclan-list


-- 
Gary Warren King
metabang.com
http://www.metabang.com/