On Oct 19, 2005, at 2:53 PM, Joost de Greef wrote:
>>> I'm trying to make a lam version that differs from
>>> the one that is installed by default on the cluster I
>>> have to use. Compiling with --program-suffix=-lam nicely
>>> adds -lam to all the executables, but for some reason lam
>>> still looks for some executables without the extension.
>>>
>>
>> LAM/MPI currently does not support the use of the program transform
>> features of configure. Unfortunately, there's no way to get rid of
>> the options (that I'm aware of), as some macro we uses requires the
>> macro that provides those configure options. To be honest, no one
>> has asked about it in the last many years, so it's never really
>> popped up on our radar to look at. I'm taking a look at supporting
>> the program name transform features in 7.1.2, but I can't make any
>> promises.
>>
> how much work would implementing (in a shortcut fashion) this
> myself be?
> it seems all the executable names are tucked away in defines
> in lam-7.1.2b27/share/include. would i get anywhere by just
> changing DEFTRESETH, DEFTHBOOT, etc?
I was just starting to look at what would be required - it'll be
slightly easier for the quick fix, since you don't have to do the
generic transform support we'll have to do in order to implement
things correctly :). The places I know you would have to modify are:
- etc/lam-conf.lamd
- share/include/*.h <-- changing all the DEFT defines
- otb/mpirun/totalview.c (in the sfh_argv_add call)
- a bunch of places in otb/sys/* where we neglected
to use the #defines
There might be some more, but those are the ones I can think of off
the top of my head.
Brian
--
Brian Barrett
LAM/MPI developer and all around nice guy
Have a LAM/MPI day: http://www.lam-mpi.org/
|
