Re: [OLPC library] [Re] 'OLPC-Health' takes off !!

["K.S. Bhaskar" <bhaskar@bhaskars.com>]

Andreas --

GT.M is bootstrapped.  Just like you need a gcc to build a gcc, only the dependency is not as strong.  Remember that GT.M is both a language and a database, and so bootstrapping is not an unreasonable requirement.

But, in any case, this is outside my expertise.  I will take this up with the developers and get back to you.  Since I am traveling, I may not close the loop till early next week.

Out of curiosity, are you compiling GT.M as a matter of principle or because the OLPC does not have an i686 instruction set?  The binaries that we provide have gone through extensive testing whereas the binaries that you build will not have the same testing.

Regards
-- Bhaskar

On Feb 7, 2008 3:16 AM, Joseph Dal Molin <dalmolin@e-cology.ca> wrote:

I am forwarding this to KS Bhaskar, product manager for GTM......I am
sure he will sort this out for you.

Joseph

Andreas Tille wrote:
> On Wed, 6 Feb 2008, Seth Woodworth wrote:
>
>
> Tried to compile gtm_V53001_linux_i686_src.tar.gz which I downloaded from
>      http://sourceforge.net/projects/fis-gtm
> but failed. :-(
>
> Readme suggests:
>
>    make -f sr_unix/comlist.mk -I./sr_unix -I./sr_linux gtm_ver=`pwd`
>
> (well - at least this is my interpretation of the very short description)
> Build process throws lots of warnings like
>
> cc1: note: obsolete option -I- used, please use -iquote instead
>
> and finally fails with
>
>   make[1]: *** No rule to make target `libmumps.a(cmerrors_ctl.o)', needed by `libmumps.a'.
>   Stop.
>
> which is not astonishing that the result of the previous process was
> just a bunch of text files in `pwd`/pro/obj with names *.d .
>
> It might be that a binary version of GT.M is needed to build the source
> which just sucks.
>
> Any ideas?
>
> Kind regards
>
>           Andreas.
>