Tancredi,
It looked to me like the clever makelist copied the descriptive file
(commis_r_pri_nots.crl) to tmp/commis_r_pri.crl, so that the internal
name would match up with what is in the msql database. That is fine.
BUT I think either the ccrl statement or the VXWORKS68_GCC statement
(the first one) is operating on commis_r_pri.crl in ~/commis/coda/crl,
not the one in tmp/. Then makelist copies the .o to
/usr/local/blast/obj/commis_r_pri_nots.o, as it should. And puts the .c
in src/commis_r_pri_nots.c. And I do the proper ln -s, so that the ROC
sees the filename it expects, but the link shows me which descriptive
file I'm really using.
If you look at src/commis_r_pri_nots.c, you will see it is IDENTICAL to
commis_r_pri.c, even though the crls are different. Or maybe look at
ts2_pri_ns.c and ts2_pri.c, since I don't know where I ended up with the
commis one.
I was trying to fix makelist the other day by moving the cd tmp around,
or inserting tmp/ in front of $DOING_FILE. ButI did something wrong.
I'll revisit it today.
Karen
Tancredi Botto wrote:
>
> Dear Karen,
>
> >
> > The clever version of makelist in ~/commis/coda/crl to preserve the
> > original descriptive source filename (e.g. commis_r_pri_nots) in the
> > resulting .o, while having it internally consistent with the standard
> > filename (e.g. commis_r_pri), has a bug in it.
> >
> > It actually precompiles the STANDARD crl (e.g. commis_r_pri) and then
> > puts the result in the descriptive filename (e.g. commis_r_pri_nots.o).
> >
>
> yes, and the .o file is put in ..../daq/objs
> The whole point is that you should step one directory up and link
> the default name to the new obj file
>
> /bin/ln -s ./objs/commis_r_whatever.o commis_r_pro.o
>
> in this way we do not need to delelte/rename executables which could havw
> been a source of confusion. Am I missing something here ?
> At any rate, you can now rename the .o file. The later finding was
> that this had to be compiled from a standard commis_r_pri.crl (e.g.)
> The name of the obj file did not matter
>
> -tb
>
> P.S.
> hoe is het met de nieuwe TS ?
>
> > I found this by looking in the ~/commis/coda/src directory at the
> > intermediate .c filename, which does NOT have the changes I made in
> > commis_r_pri_nots.
> >
> > Until I can figure out enough about Linux to get makelist right, or
> > Tancredi logs in from Slovenia, we are stuck with the standard files. I
> > can also cheat things to change the readout list in a backdoor way.
> >
> > Karen Dow
> >
This archive was generated by hypermail 2.1.2 : Mon Feb 24 2014 - 14:07:28 EST