Recent changes to bugs

#26 segfault when create thread in plugin(perl)

Eduardo M Kalinowski — Sat, 18 Jun 2022 12:01:41 -0000

status: open --> closed-fixed
Group: --> v1.0 (example)

#26 segfault when create thread in plugin(perl)

Eduardo M Kalinowski — Sat, 18 Jun 2022 12:01:25 -0000

Even though I didn't do anything specific to treat this bug, your code now works, and I was able to confirm it is running in the background. Maybe changes in libperl or the fact that now KC itself is multi-threaded.

That said, starting threads in Perl is something that has not been test, and there might be unexpected effects, including crashes. So I'd call it an unsupported feature.

#32 Broken $::world->trigger() calls (part 2)

Eduardo M Kalinowski — Sat, 18 Jun 2022 11:53:26 -0000

status: open-accepted --> closed-fixed
Group: --> v1.0 (example)

#32 Broken $::world->trigger() calls (part 2)

Eduardo M Kalinowski — Sat, 18 Jun 2022 11:53:13 -0000

Since the encoding pragma does not exist anymore, the code in ex6.pl now works as expected.

#34 Crahes and patches

Eduardo M Kalinowski — Sat, 18 Jun 2022 11:45:44 -0000

status: open --> closed-fixed
Group: --> v1.0 (example)

#40 Segmentation fault on $::world->alias call

Eduardo M Kalinowski — Sat, 18 Jun 2022 11:42:56 -0000

status: open --> closed-works-for-me
Group: --> v1.0 (example)

#43 Trigger causes text to display out of order

Eduardo M Kalinowski — Sat, 18 Jun 2022 11:40:28 -0000

status: open --> wont-fix

#43 Trigger causes text to display out of order

Eduardo M Kalinowski — Sat, 18 Jun 2022 11:40:13 -0000

I've looked again at this issue and, as mentioned before, the problem is that channels adds a trigger that captures every line, suppresses normal processing, and feeds it to a Perl function (that in most cases prints the line). Your trigger is run before the channels trigger, and it only changes the styling of the line, which is printed using the regular line-processing code (that is, receive the line and print it.)

(The fact that it's a highlight is not important - the same problem can be seen with a trigger that sends something back to the server, or even a trigger that uses Perl, as long as it's not a gag trigger.)

Some years ago some big changes were made to make KildClient more responsive by using multiple threads. This unfortunately can add some concurrency problems such as this one. Not having to use Perl as a trigger action means that the line matched by your trigger finishes processing earlier and is printed out of order.

Unfortunately changing that would require some very extensive and complicated changes, so I'm marking this as won't fix. But there is a workaround, even if you need the channels plugin or a similar functionality : if you use Perl in your trigger, it'll use the same processing pathway as the other lines, and everything will be in proper order again.

#45 SB TERMTYPE answer followed by NUL byte

Eduardo M Kalinowski — Fri, 10 Jun 2022 17:29:21 -0000

status: open --> closed-fixed
assigned_to: Eduardo M Kalinowski

#45 SB TERMTYPE answer followed by NUL byte

Eduardo M Kalinowski — Fri, 10 Jun 2022 17:28:59 -0000

Fixed in b8978e0