Which ifc features not yet implemented in ifort and when will they be?

[dfumento]

Hi,
Apparently that -zero flag is not yet implement and neither is -CU or any of the -C flags. What else is not yet implmented in ifort and when will they be implemented? Can we please get a complete list?

Thanks in advance,
David

[Steven_L_Intel1]

-zero should be available in the next update. The various -C options are planned for a future release. The only other unimplemented command line option I know of is -C90, and that's not going to return.

[dfumento]

Thank you.

[timconnors]

I suppose that with the complete runtime check options, we get proper debug info at crash time as well?

Currently (20040412Z Package ID: l_fc_pc_8.0.046), I get something similar to:

ifort -o xyconvlbg xyconvlb.go bits.go bitsc.o  convertlbxy.go -u 
-nbs -tpp7 -axW -fno-alias -ansi_alias -ip -align -warn all -O0 -C  
-L/home/office/tconnors/software/Linux-i386/lib -lpgplot 
-L/usr/X11R6/lib -lX11 -lpng -lcfitsio 
forrtl: severe (64): input conversion error, unit 20, file 
/nfs/cluster/cosmic2/tconnors/magellanic/hvcsim/output/lb.dat
Image              PC        Routine            Line        Source             
xyconvlbg          080B646B  Unknown               Unknown  Unknown
xyconvlbg          08089593  Unknown               Unknown  Unknown
xyconvlbg          08089A3A  Unknown               Unknown  Unknown
xyconvlbg          0809DA0F  Unknown               Unknown  Unknown
xyconvlbg          0809CA86  Unknown               Unknown  Unknown
xyconvlbg          0804F275  Unknown               Unknown  Unknown
xyconvlbg          0804D413  Unknown               Unknown  Unknown
xyconvlbg          0804A2C8  Unknown               Unknown  Unknown
xyconvlbg          402688AE  Unknown               Unknown  Unknown
xyconvlbg          0804A181  Unknown               Unknown  Unknown
forrtl: error (76): IOT trap signal

(You can see the compiler flags I used - which include -g and -C)

Ideally, it would behave like ifc 7.0 - it would tell me the line number of the input file it crashed at (in this case, it was the ~20000th line of the file, where I started getting -INFs in my input, and read (20,*) doesn't know what to do about them), it would tell me the line number of the code it crashed at, etc. Indeed, the output should not be dependant upon the -C runtime checks -- they should just depend on the -g debugging output.

Perhaps I have done something wrong? Some of those flags came from the ifc 7.0 installation, and don't seem to be documented in the ifort manpages, but they don't crash anything, and when removed, the output is still the same.

[Steven_L_Intel1]

I have seen some of these sorts of errors give additional info on the file. This particular case would not be a -C type check, by the way.

Have you tried it with -g?

[davidham]

You need the -traceback option to put line number information in the stack trace. This is quite separate from the -g option.

[timconnors]

Aha. -traceback is good. Thanks.