Ok, changes pushed. I believe I've a working update for jtxplus, and I believe it's portable to all supported architectures.
I'll keep an eye out for issues. If all goes well, I intend to start working through some of the other X primitives (and, eventually, the RAT primitives). If I have to hold off at some point for release management purposes, I think that that should be fine - this effort can be completed in J9.6 or whenever. Thanks, -- Raul On Sun, Dec 3, 2023 at 5:37 PM Henry Rich <[email protected]> wrote: > > It is OK to get this out whenever you have it ready. I don't expect any > changes that would interfere with your work. > > hhr > > On 12/3/2023 2:59 PM, Raul Miller wrote: > > I've found and addressed the problems raised in this thread. > > > > It boiled down to two issues: > > > > (1) I misunderstood an issue related to memory management. This is > > resolved and should not recur. > > > > (2) On windows, we're using mpir, which is an ancient fork of libgmp. > > This is a fixed issue for addition, but may hint at future problems, > > where I'll have to re-implement missing function(s). > > > > However, these problems of mine have dropped this update out of the > > j9.5 release plan. So, for now at least, I'll plan on plodding along > > in a separate branch (which will need to track the master branch for > > those changes). > > > > Thanks, > > > > -- > > Raul > > > > > > On Sat, Dec 2, 2023 at 5:51 PM Raul Miller <[email protected]> wrote: > >> (0) Yes, it looks like I had already taken steps to deal with the 0x > >> case, and clearly they were inadequate. > >> > >> (1) I cannot find the message you are referring to. > >> > >> (2) I am plugging away, trying various approaches, maybe I'll have > >> this fixed soon. (It would be faster if script/buildga.sh could be > >> told to build only one instance of j, rather than all the various 32 > >> and 64 bit instances. It would also be faster if I did not allow > >> myself to be distracted while waiting for a build to complete.) > >> > >> Thanks, > >> > >> -- > >> Raul > >> > >> On Sat, Dec 2, 2023 at 11:25 AM Henry Rich <[email protected]> wrote: > >>> A couple of messages ago I pointed to the line > >>> > >>> if (unlikely(!n)) {fr(z); R X0;} /* this X0 is presumably the one > >>> that happens */ > >>> > >>> as being suspect, because fr should never be called explicitly. > >>> > >>> That message contained an explanatory paragraph at its end. > >>> > >>> hhr > >>> > >>> On 12/2/2023 2:57 AM, Raul Miller wrote: > >>>> That sounds very plausible. Thanks. > >>>> > >>>> Previously I had a special value that I always used for 0x. It makes > >>>> sense to be using it here. > >>>> > >>>> I'll try that and see if it clears things up. > >>>> > >>>> Thanks again, > >>>> > >>> ---------------------------------------------------------------------- > >>> For information about J forums see http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- > > For information about J forums see http://www.jsoftware.com/forums.htm > > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
