Adapting software for LFS
 

Hello everybody!

Its great that BLFS comes with instructions for installing lots of stuff that will probably be needed sooner or later by most people. But what if I want to install something that is not in this list?

I would have to figure out:

What dependencies does this specific package have? I have tried reading the README and checking the websites for some software, but usually this information is hard to come by. Is there some way to find out without digging through every include in the source files? I can run ./configure and see what is missing, but it will only tell me the first package that is missing, then exit. There is no way to know in advance exacly how many packages i would need to install without actually doing it, is there (if not listed on the BLFS web, of course)?

I have run gentoo before, and the emerge command solves these problems, but without letting me know how its done. I suppose the developers keep a list of dependencies manually for emerge? Or maybe there is some kind of a script I can run that passes through the source files, letting me know what other packages are needed (judging from the include files)...?

Christer, Sweden

we could setup a wiki to handle this.
if somebody goes throught the step-by-step proccess of installing something without know its dependencies. they can list all of the packages that was installed.

Even if you had some dependencies, somebody else might find they dont have them, and then just add those extra dependencies that you failed to notice as a result of a pre-resolved dependancy.

And with the new linux from scratch site up, I believe there is a wiki that I never noticed before, perhaps they already have something like this.

We'll both have to take a look at: http://wiki.linuxfromscratch.org/

Yes, thats a good idea.

But i wish there was a simpler way to find out about the dependencies though. A wiki is good, but the information may not be 100% accurate depending on software version, host system, etc. And there are often switches you can use during install, some of which will give you additional dependencies for the extra features.

It should be possible to write a script that:

1) For every source file, read the #includes
2) Check if the #include exists in the path on the local system

If it doesnt, it would use a database to look up the header file and check what package it is a part of. It would check if this new package has dependencies and tell the user about those as well.

Of course this requires a updated database and contribution by users to make it work. Maybe it would be simpler to just let the script read what dynamic libraries are used after the package has been installed and post this to a database. But this could vary between different host systems and installation switches used. And some packages are built with static libraries...

Maybe someone with more linux experience has a better solution....