This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Fwd: Question on PR25289]


Thomas Koenig wrote:
We have one question. What, if any, problems do we have on a system that does not support GFC_INTEGER_8? Are there any such systems?

I don't think there are any systems without GFC_INTEGER_8. Can somebody else confirm this?

Do you see any issues with the record_marker size options?

No problem. I just wonder what will happen with large files on systems where sizeof(off_t) is 4. I can't test this, however.

Right now in the 4.1 branch and trunk, large file attempts yield an error on record number because the frontend is using GFC_INTEGER_4. The patch simply extends that to kind=8 where it exists.

--- 129,143 ----
kind = bitsize / 8;
if (kind == 4)
! {
! saw_i4 = true;
! gfc_record_int_kind = 4;
! }
if (kind == 8)
! {
! saw_i8 = true;
! gfc_record_int_kind = 8;
! }

This only works if the code is called with kind=4 first, with kind=8 second. Is this guaranteed?


The following minor update will avoid any concern here.


+   if (saw_i8)
+     gfc_record_int_kind = 8;
+   else
+     gfc_record_int_kind = 4;

If GFC_INTEGER_8 is indeed guaranteed to be present, then I think
we should bother with this.  Just put in assert(HAVE_GFC_INTEGER_8)
somewhere.


At this point I propose we commit the patch to 4.2 and allow some exposure to see if there are any real problems. We can then always fix or revert if something shows up. Is this not why trunk is not a release yet?


OK to commit to 4.2?

Jerry



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]