nil deref error in treemacs when explain-pause-mode is active (root cause: process-filter wrap nil callback) about explain-pause-mode HOT 5 CLOSED

Hi!! Thanks for the bug report. I'm 100% sure it's because I didn't read the documentation carefully enough for set-process-filter. Quote

set-process-filter process filter
This function gives process the filter function filter. If filter is nil, it gives the process the default filter, which inserts the process output into the process buffer.

Well, duh, filter is nil here, because treemacs wanted it to go into some buffer, it must have done something like this:

(set-process-filter my-process nil)

This would drop into https://github.com/lastquestion/explain-pause-mode/blob/master/explain-pause-mode.el#L1878 which drops into https://github.com/lastquestion/explain-pause-mode/blob/master/explain-pause-mode.el#L1838 which drops into https://github.com/lastquestion/explain-pause-mode/blob/master/explain-pause-mode.el#L1807, which tries to call apply nil.

I'll pull a repro case together. The fix needs to be around https://github.com/lastquestion/explain-pause-mode/blob/master/explain-pause-mode.el#L1874 and for set-process-sentinel, too.

I think this is an easy fix, it's possible to advise the code to fix it locally too but it's a bit of a pain. I'll just suggest waiting, I'll get this fixed today.

Thanks again for reporting!! I'm surprised no one else has hit this... kind of a glaring bug 🤣

from explain-pause-mode.

Repro case

(setq proc (make-process
            :name "test"
            :buffer "test"
            :command '("bash")))
(set-process-filter proc nil)
(process-send-string proc "ls -al\n")

from explain-pause-mode.

Hi!! Sorry this took longer to fix then I thought. This should fix the issue, please re-open if it doesn't!

from explain-pause-mode.

Hi @tomfitzhenry I was going through old PRs to add repro cases to a set of integration test - a little creepy but I saw your reference 😀

with-editor is in magit and is relatively complex, the other open issue #26 is also related to with-editor. I have (I think) finally quashed #26 in the open PR #42. The bug you hit is likely to be related to that, what happened is probably the filter intercept code failed with an error and so never returned the value back out to with-editor. When #42 is merged I give it more then even odds that the bug you found will be fixed, but if not, feel free to open a bug with just very basic info, it's super appreciated if you can debug into it but not necessary :_)

from explain-pause-mode.

Hi @tomfitzhenry I was going through old PRs to add repro cases to a set of integration test - a little creepy but I saw your reference grinning

Not creepy at all! That's one of the handy parts of Github.

When #42 is merged I give it more then even odds that the bug you found will be fixed, but if not, feel free to open a bug with just very basic info, it's super appreciated if you can debug into it but not necessary :_)

I've now raised a bug at #46 .

Thanks for your great work!

from explain-pause-mode.