[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: building vanilla

From: Yavor Doganov
Subject: Re: building vanilla
Date: Thu, 27 Dec 2007 22:12:18 +0200
User-agent: Wanderlust/2.15.1 (Almost Unreal) SEMI/1.14.6 (Maruoka) FLIM/1.14.7 (Sanjō) APEL/10.6 Emacs/22.1 (i486-pc-linux-gnu) MULE/5.0 (SAKAKI) (gNewSense GNU/Linux)

Калоян Доганов wrote:
> What is the GNU Build System?  GNU Make?

Autoconf, Automake and Libtool.  A typical makefile that conforms to
the GNU standards behaves as I described.  Anyway, this is way off
topic (my fault) -- during the past days I just tried to find out the
reason for this behaviour and it makes a lot of sense when you think

I can give a lot of examples, unfortunately most of them Debian

> Those are nice features, but it's not obvious to me how they will be
> implemented right now. 

It is almost certain that we will make a mistake if we start thinking
about all of them now in detail.

Perhaps we see the things differently.  For me, the cron job is just a
way to automate the "build run" and do it periodically.  There should
be no difference compared to simple manual cvs update && make && cvs

> Just listing them does not add an argument against rebuild and
> add/commit decoupled in two stages

Right.  Let me present it in a different way.

Imagine a simple C program.  We list the source files in the makefile,
so that make does know what to compile.  But we don't link them at the
end, because, say, the main purpose of this test program is to rebuild
the object files; linking is just a nice feature.  We decide to run
the link phase at the end from within a script that acts like a build
bot and to find (determine) the necessary object files, we run other
tools like find.

Does it make sense?  Absolutely not -- make surely knows better what
objects to link even if it is not an implicit rule.  So I don't see
how we can complicate the build process if we run the linking commands
only conditionally.  Make still has all the necessary information and
acts when we say so, and acts with no or little chance for a serious

> It is not dangerous to try it.

Yes, wait to see the exact implementation.
Too much theory can be confusing :-)

reply via email to

[Prev in Thread] Current Thread [Next in Thread]