verilogams moved invocation directory

hi, i moved the location i start cadence/verilogAms and now i get
errors like:

ncelab: *F,DLIORG: Library (physical path) /dsk/analoglogic/p1/Project/
design/dfII/analogLogic was originally compiled with logical name
'analogLogic'. Please include the original definition of library in
cds.lib file.

i've deleted the .pak , inca.* and .inca* files from my library area.
what else do i need to delete to remove these hardcoded paths from my
simulation area?

 

stroller wrote, on 03/31/08 20:43:

Normally this occurs when you rename a library. You need to delete the pak files
from the library itself (not just the simulation directory), but you probably
also should delete the simulation tree. If you have an ASSIGN TMP dir, you
should delete the pak files from there too.

It's not the absolute path that matters - it's the name of the library in the
cds.lib.

There's a CCR to get this to happen automatically - i.e. get it to clean up
properly rather than force you to delete the files by hand.

Regards,

Andrew.

 

Thx!

 

Andrew, thanks for the post... I have seen this problem myself. If I
set compile all and elaborate all to true, shouldn't this get rid of
the inca and pak file issues?

The part of your post I don't quite follow is the part about cds.lib.
for example if my old cds.lib was

DEFINE foo /tmp/dsk/path/foo

and my new cds.lib is

DEFINE foo /tmp/dsk/newpath/foo
DEFINE foo_old /tmp/dsk/path/foo

Why does the simulator care that foo used to be /tmp/dsk/path/foo?
What is it that is looking in cds.lib and remembers what it used to
be?
Or is it that foo_old has the path that foo used to have... again what
in the simulator noticing this and why doesn't recompiling fix it?

thanx

David

 

DReynolds wrote, on 04/03/08 12:45:

I'm surprised. It's normally the library name that's the problem, not the path.
That's because the library name is stored within the pak file.

Note that a re-netlist and compile will not remove the pak file within the
library - it only affects the pak file underneath the ADE netlist directory.
Unfortunately this is (if my memory is correct) seeded from the pak file in the
library, so if that's messed up, the mess-up propagates.

Anyway, the workaround is to manually delete the pak file, and hopefully we'll
fix things so that this gets done behind the scenes, or it is at least smart
enough to ignore the duff pak file.

Regards,

Andrew.