[Gdal-dev] GDAL 1.2.4 RC1 available for testing.

[Alessandro Amici]

Frank,

first, i wrote 1.6.0 from memory and i got i wrong. the soversion of gdal 
1.2.4 is actually 1.5.0.

On Saturday 06 November 2004 17:55, Frank Warmerdam wrote:
> My understanding is that libtool is using -soname and other mechanisms so
> the versioning information serves to select alternate compatible versions
> automatically, but that this isn't true of the shared libraries I create
> without libtool.  So my thinking has been I might as well use an
> understandable name for the non-libtool builds since it doesn't serve much
> other purpose.
>
> It also makes it somewhat easier to identify which kind of builds was
> used for a given GDAL build.

i understand your reasoning, but, as a user, i'd find rather confusing to have 
two different filenames for the same version of the same library, where the 
only difference is an obscure build-time option.

should i myself maintain a shared library i would probably use the libtool 
versioning convention and call the current iteration of gdal version 1.5.0, 
thus avoiding the name conflict.

> Do you see any compelling reason to use the same naming convention for the
> .so files?

this might be considered mainly a matter of taste (and i know your dislike for 
version number inflation ;-) ). so i'll leave it to your call.

> > BTW: the libtool build survived a make -j 200 ;)
>
> Cool, I would imagine there is a law of diminishing returns on doing so
> many at once though.

not on linux 2.6 with the new O(1) scheduler (and 1.2 gig of ram) :-). on my 
machine i did some tests a few months ago. if i recall correctly gdal compile 
time goes down from -j 1 to -j 4 or 5 and then stays flat up to -j ~250 where 
the machine starts swapping. pretty impressive :)

cheers,
alessandro