Weird handling of Ctrl+C

[pkoch]

When I add the following to my init.pl:

:- set_prolog_flag(readline, readline).                                        
[library(readline)].

readline does get used. However, pressing Ctrl+C doesn't make the shell immediately react. I have to press it again, and then some other key, leaving me with the message:

?- asdas<press Ctrl+C twice, then another key (like an arrow)>
WARNING: By typing Control-C twice, you have forced an asynchronous
WARNING: interrupt.  Your only SAFE operations are: c(ontinue), p(id),
WARNING: s(stack) and e(xit).  Notably a(abort) often works, but
WARNING: leaves the system in an UNSTABLE state

Action (h for help) ? Unknown option (h for help)
Action (h for help) ? 

Did I mess something up? What can I do to move this closer to the current behaviour when using editline? Would a PR fixing this be accepted?

[JanWielemaker]

Only set the flag, don't load the library. The startup sequence will do that if the session is interactive.

^C while editing doesn't seem to work. I'm happy to get a PR that fixes this. Otherwise I consider it low priority as there is no point to interrupt editing and the libedit command line editor is the default anyway.

[pkoch]

Thank you for your reply and tips! It feels great to be addressed! :)

Do you have any pointers for where I should start looking to fix that? Are there perhaps examples I can take a look (how is this handled with libedit)? Any inspiration would be appreciated.

[JanWielemaker]

The sources are in packages/readline. Libedit won't help you much. The APIs of the two command line editors are quite different. IMO the readline one carries a bit too much history, but true, GNU readline is a more capable editor. libedit can be programmed from Prolog though and can provide command line editing from multiple inputs, for example when using the ssh package.

Fixing will require some knowledge about C and Unix signal handling ...

[pkoch]

Pretty comfortable with C and Unix. I'm not familiar with swipl (handling of signals, building it, etc).

I'm guessing I should start by studying this part of the sources?

packages-readline/readline4pl.c

Line 229 in 24f2dad

Signal handling wrapper.

[JanWielemaker]

That is a good start. I'd build the system for debugging so you can see what happens and where the ^C ends up. It is not very hard. See https://www.swi-prolog.org/build/unix.html and CMAKE.md in the topdir (the last fro setting up the debug build).