bug#51026: 29.0.50; Edebug leaves data in symbols plist after instrumentation

[Arthur Miller]

Lars Ingebrigtsen <larsi@gnus.org> writes:

> Arthur Miller <arthur.miller@live.com> writes:
>
>> +(defun edebug--strip-plist (symbol)
>> +  "Remove edebug related properties from SYMBOL's plist."
>> +  (dolist (prop '(edebug edebug-behavior edebug-coverage
>> +                         edebug-form-spec edebug-freq-count
>> +                         ghost-edebug))
>> +    (cl-remprop symbol prop)))
>
> This will break edebug -- edebug-form-spec is set when loading
> edebug.el, it's not something that happens when you instrument a
> function.

Ahh that pcase in the niddle of nowhere; what does it do there? Why is it not in
some "init" function? :-)

Yes, it is true what you say, but only for a handful of symbols, not even all of
those handled in that pcase. I think it is emitted only for let, let*, setq and
quote, but it is not important. See attached patch, if it is acceptable. It will
check if a symbol is one of those specially treated and do nothing for those.

I think that is what you mean? I don't see any other place that adds properties
and I don't see any edebug properties in "random" symbols unless instrumented.

> I don't know about the other symbols.

That seems to be the only one added automatically.

0001-Clean-edebug-props-on-instrumentation-removal.patch
Description: Text Data