Andre Poenitz <[EMAIL PROTECTED]> writes:
| On Thu, Mar 29, 2007 at 03:23:08PM +0200, Lars Gullik Bjønnes wrote:
| > Jean-Marc Lasgouttes <[EMAIL PROTECTED]> writes:
| >
| > | >>>>> "Andre" == Andre Poenitz <[EMAIL PROTECTED]> writes:
| > |
| > | Andre> On Sun, Mar 25, 2007 at 10:31:24PM +0200, Jean-Marc Lasgouttes
| > | Andre> wrote:
| > | >> >>>>> "Andre" == Andre Poenitz
| > | >> <[EMAIL PROTECTED]> writes:
| > | >>
| > | Andre> libtool is a waste of time in the year 2007. That's certainly
| > | Andre> just my opinion, but anyway. I also believe the kind of
| > | Andre> variations between different version of Unix leading to and
| > | Andre> rectifying the existance of autotools are gone nowadays (with
| > | Andre> Linux as the only surviving variety, the rest is dead even if
| > | Andre> the coroner's still out in some cases).
| > | >> I think the code we had in the pre-libtool era was perfectly
| > | >> suitable.
| > |
| > | Andre> Suitable for what?
| > |
| > | Compiling LyX.
| > |
| > | Lars' big reason for using libtool, AFAIR, was that that would allow
| > | us to build parts of LyX as shared libraries. In practice, we have
| > | never done that.
| >
| > Currently we can only flip a switch if we want to.
| >
| > If we stop using recursive make (and automake supports this) we might
| > be able to do without the convenience libs.
|
| I would like to see real numbers from this direction.
I don't have numbers from our project. But on one project at work
(thousands of C/C++ files) use some 1.30 min on a null build (recursive
make), with the
new system that we are implementing (non-recursive, single
makefile) this goes down to 10-15 seconds,and that include parsing and
generaton of the single makefile each time.
So imho this is the direction where we should put our effort. Not
conveing some kind of lumping/merging og writing overly large source
files.
| I think getting rid of the recursive make would already be beneficial.
By quite a lot I would guess.
--
Lgb
