@alexshendi Did you start from the port? It has a number of patches that were probably never upstreamed.
I can look into updating the port soon.
@bcallah Yes I did, but I must have done something wrong. I can build an apl executable, but it segfaults.
@alexshendi Send me what you have.
I currently have some problems sending emails. Could you please notify me,
if you have received anything.
Thanks in advance.
@alexshendi I haven't received anything.
Please check again now. I have had problems with SMTP auth. But it seems to have worked now.
@alexshendi Still nothing. Just post a diff to ix.io or github or something.
@alexshendi Oh, I see what you did here. No, instead run `make update-patches' in /usr/ports/lang/apl then `cvs diff -uNp' and send me that diff.
A couple of questions :
- How should I fetch the ports tree? By ftp from cdn.openbsd.org or via AnonCVS?
- After I have obtained the ports tree, I run "cvs diff -uNp" in my source directory? How does cvs know which port to check?
Thank you very much for explaining.
$ cvs -email@example.com:/cvs checkout -P ports
Then you go to the port directory itself, i.e. lang/apl in your case and run your cvs commands. CVS is smart enough to do the right thing.
What I still need to know:
When I checkout ports lang/apl contains a port set up for GNU APL 1.7. Do I then update it to match my source directory i.e. ~/sources/packages/apl-1.8p2?
Sorry, if I'm dumb, but I have never done this before.
Again, thank you for your patience.
@bcallah Currently downloading the ports tree.
@alexshendi Honestly I would just tackle the bug report right now. I already finished the update but a fix for that compile error is very necessary. Your workaround doesn't actually work.
@alexshendi It's in clang. So it may well be a general clang issue.
Some more findings:
The same problem occurs when you use g++ 8.3 from ports to compile APL 1.8. So I suspect it's not a clang issue.
Jürgen Sauermann (the author of GNU APL) has responded to my message to bug-apl. He said that the problem is already fixed in SVN. Apparently someone had similar problems under Cygwin. But the fix is the same thing that I did - - namely check if _B and _AB are defined and if so undefine them. Only in Performance.def not in Performance.cc.
I am not sure how to deal with this.
@bcallah Bug report sent.
@alexshendi Taking a look at this, you should report your issues with Performance.cc to upstream. There's no way your patch to that is OK.
This Mastodon instance is for people interested in technology. Discussions aren't limited to technology, because tech folks shouldn't be limited to technology either!