Unfortunately it's a bit more compilicated.
I could say that the dependency is *listed* in a Makefile, but it is used
for more than that.
I am trying to implement a config mgt tool that is modules-aware.
I started with mpkg, but we decided to move away from it because:
- We can't contribute to or track any ongoing development as it uses AFS (a
no-no from work)
- The GNU .mk files are not as robust or featured as we need
I am now adding modules support to the stuff from www.openpackages.org, but
it is going very slowly. My original work was done to the V1 OPmake .mk
files (which were hacked from NetBSD, as I recall). V2 of OP will probably
use DarwinPorts, but again, that system is still evolving and my management
does not want me to pursue that course at the moment.
I am currently usinog opmake and a set of hacked .mk files from FreeBSD'd
port/ tree, which has been modified to ignore their packaging and do
mpkg-style installs.
I need to layer in the dependency stuff, which will need to handle
dependencies for:
- bootstrap
- build-time
- run-time
mpkg (at least the version I have, the tarball from about a year ago, as I
recall) speciifies dependency targets that are suitable fas arguments to
"module add", which means one can specify a particular version or not (and
get the default version).
As this system could eventually have thousands of packages that will each
have their own dependencies, I'd really like to be able to specify a range
of versions for dependent packages.
And while email is nice (I gather there is an archive somewhere), at some
point I think it would be really useful to distill this information in a
TWiki. Is there a machine the modules group has access to that could run a
TWiki (basically a web server)?
H
--
> what is enforcing the dependency checking?
> ...
> > Any ideas on how one might implement "version ranges" for dependency
> > checking code that uses modules?
> >
> > Something like:
> >
> > gmake>=3.78
|