Re: RFC: toolenv - for dealing w/ tools' environments

Jonathan H wrote:

> I was wondering why you'd have the one-line shell scripts
> to invoke the real executables via toolenv. Surely there
> would be a requirement to be able to invoke the same
> executable in different environments depending on what
> tool packages it is being used with?

(Sorry to be slow, as well.)

If tools need their environments intertwingled, it's a
fairly static thing.  For Specman, for example, you'd set up

   toolenv specman,vcs,ncverilog,simmagic specman ...

where specman is itself, and vcs/ncverilog/simmagic are
simulators that you have and which specman needs to work
with.  You might occasionally get a new simulator and need
to re-jig the above, but you would be unlikely to suddenly
want to do

   toolenv specman,xemacs,nessus specman ...

for (outrageous) example.

> I was thinking that users would need to "declare" what
> environment they need and then just invoke the executables
> directly. Something like:
> 
>     % toolenv verisity-specman cadence-LDV
>     % specview --special-option=on -c config1 mumble.spec

I don't like this, because it means (for instance) that some
weird-and-wacky LD_LIBRARY_PATH (and they all are... :-) is
now floating around.  But reasonable minds can differ on
this, for sure.

> I guess I'm thinking along the lines of my old
> 'usepackage', but I never allowed for the configurations
> to vary depending on what other packages were requested or
> already in use.

I regret to say my fading brain cells had forgotten about
usepackage -- which is indeed probably where I should've
started.  I will give it a good write-up in the forthcoming
"documentation".

Will


-------------------------------------------------------
This SF.net email is sponsored by: Etnus, makers of TotalView, The debugger 
for complex code. Debugging C/C++ programs can leave you feeling lost and 
disoriented. TotalView can help you find your way. Available on major UNIX 
and Linux platforms. Try it free. www.etnus.com