wrong-number-of-arguments in post-command-hook

wrong-number-of-arguments in post-command-hook

[ndame]

[-- Attachment #1: Type: text/plain, Size: 662 bytes --]

I'm working on some code using post command hook and there were some errors
in it which were not trivial to find, because post-command-hook does not print
backtraces when debug on error is enabled.

It prints an error message though, here are two examples from my code:

Error in post-command-hook (my-post-command): (wrong-number-of-arguments (2 . 2) 1)

Error in post-command-hook (my-post-command): (wrong-type-argument listp t)


Seeing these I was wondering why the error does not contain the called function name.
Is that info not available? It could help debugging a lot if the error also printed the 
function name alongside with the error.

[-- Attachment #2: Type: text/html, Size: 735 bytes --]

Re: wrong-number-of-arguments in post-command-hook

[Stefan Monnier]

> Error in post-command-hook (my-post-command): (wrong-type-argument listp t)
>
> Seeing these I was wondering why the error does not contain the called function name.
> Is that info not available?

More or less: it's in the stack somewhere, but it's not trivial to
extract that info reliably.  The code that runs the hook does
store&print the name of the top-level function called
(`my-post-command` here) to try and help a bit.

> It could help debugging a lot if the error also printed the 
> function name alongside with the error.

Have you tried `M-: (my-post-command) RET`?


        Stefan

Re: wrong-number-of-arguments in post-command-hook

[ndame]

[-- Attachment #1: Type: text/plain, Size: 120 bytes --]

> Have you tried `M-: (my-post-command) RET`?

For some reason this radical approach didn't even occur to me. :) Thanks.

[-- Attachment #2: Type: text/html, Size: 140 bytes --]

Re: wrong-number-of-arguments in post-command-hook

[Stefan Monnier]

>> Have you tried `M-: (my-post-command) RET`?
> For some reason this radical approach didn't even occur to me. :) Thanks.

Of course, you shouldn't need that: the normal debugger should kick in
"the normal way".  You might want to file a bug report.


        Stefan