Re: conditionally using libtool

[Russ Allbery]

Ed Hartnett <address@hidden> writes:

> Sorry, I don't follow here.

> Firstly, I recently switched a large package to libtool, and it didn't
> seem to slow down the build all that much. It's true that it makes the
> build output harder to read, but that's no big deal.

On a modern Linux box with a decent shell, you don't really notice.  On
older Solaris and some more obscure systems (we had a developer using
UX/4800 R11), it would slow down the build by a factor of three or more.

> Secondly, the build is something that the user does exactly one time,
> so what difference does it make if it takes longer? It's a one-time
> cost.

Uh, I actually develop my software, not just use it.  :)

> Naturally we all want things to be as fast as possible, but it seems
> quite excessive to maintain two different build methods for a minor gain
> in build speed in one of them.

But I don't maintain two build systems, see.  Autoconf is really cool that
way.  I have a little bit of logic in my configure.ac file and a little
bit of logic in my global Makefile and everything else uses exactly the
same make variables and just works.

> Buy a faster workstation instead, it's a lot cheaper than extra
> configuration maintenance.

Thanks, I always like people to tell me what my development priorities
could be in the absence of factual information about how much effort my
solution is.  :)

-- 
Russ Allbery (address@hidden)             <http://www.eyrie.org/~eagle/>