yeah,
i agree that moving to gcc-3.2 will cause a lot of problems, and
should be carefully coordinate with ernie bisson. (like linking with
system/external/epics libaries). for example i was unable to compile
root with gcc-2.96 installed as a compabibility package.
however, it would be nice to have a machine with 3.2 installed to iron
out all of these differences. i've got BlastLib2 and blastmc running on
my maching with 3.2.
--chris
Tancredi Botto wrote:
>To have only gcc-3.2 on the spuds is not a good idea until we also have
>it on all blast machines including dblast. It was mentioned that gcc-3.2
>was more compliant but I was not aware of any decision. It is not a personal
>issue but as you know Root is definitely not compliant and we rely heavily
>on it. I don't see the rush, given also the points below:
>
>_ Ernie bisson must be made aware so that once we are happy with gcc-3.2
> we can schedule an installation on all other machine. Note the word
> schedule ! He very much prefers that way, but so do we!! think of all
> the various root versions here and there not so long ago.
>
>- Not everybody everywhere uses gcc-3.2. At least we should provide a way
> to have a way to have two different compilations (src trees) and not
> break the experiment.
>
>- gcc-3.2 won't speed up the code or fix bugs in the existing code. We are
> supposed to use the existing code for an aymmetry measurament next week...
>
>- It will be wonderful to have gcc-3.2 for the v3 release, which already
> is not "backward" compatible.
>
>- I have no problems if you install gcc-3.2 as an alt compiler on the
> spuds. I have problems if we can't compile with gcc-2.96 anymore on the
> spuds.
>
>
>
>
>
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:29 EST