bug#3310: condition-case-no-debug

bug#3310: condition-case-no-debug

[Johan =?UTF-8?Q?Bockg=C3=A5rd]

condition-case-no-debug is not documented, either in the manual or in
NEWS. Also, there's another feature that does the same thing, except
better (if uglier):

    (condition-case nil (foo bar) ((debug error) nil))

This handles errors but gives the debugger a chance to run first.
condition-case-no-debug OTOH either handles errors, or lets the debugger
run but does not handle errors, depending on the value of debug-on-error.

If condition-case-no-debug should be kept, at least make it expand into
the above construct (and document it).

bug#3310: condition-case-no-debug

[Stefan Monnier]

>     (condition-case nil (foo bar) ((debug error) nil))
> This handles errors but gives the debugger a chance to run first.

Yes, I hate this feature.

> If condition-case-no-debug should be kept, at least make it expand into
> the above construct (and document it).

So, clearly I don't want condition-case-no-debug to expand into it.


        Stefan

bug#3310: condition-case-no-debug

[Richard M Stallman]

    >     (condition-case nil (foo bar) ((debug error) nil))
    > This handles errors but gives the debugger a chance to run first.

    Yes, I hate this feature.

You don't have to use it, so why does its existence bother you?

I agree that condition-case-no-debug seems redundant.
The other feature is more flexible and more elegant.

bug#3310: condition-case-no-debug

[Stefan Monnier]

>> (condition-case nil (foo bar) ((debug error) nil))
>> This handles errors but gives the debugger a chance to run first.
>     Yes, I hate this feature.
> You don't have to use it, so why does its existence bother you?

Exactly because I have to pay for it even if I never use it.


        Stefan

bug#3310: condition-case-no-debug

[Richard M Stallman]

    >> (condition-case nil (foo bar) ((debug error) nil))
    >> This handles errors but gives the debugger a chance to run first.
    >     Yes, I hate this feature.
    > You don't have to use it, so why does its existence bother you?

    Exactly because I have to pay for it even if I never use it.

I do not follow you.  I don't think this will bother you at all
if you don't use it.

bug#3310: condition-case-no-debug

[Stefan Monnier]

>>> (condition-case nil (foo bar) ((debug error) nil))
>>> This handles errors but gives the debugger a chance to run first.
>> Yes, I hate this feature.
>> You don't have to use it, so why does its existence bother you?

>     Exactly because I have to pay for it even if I never use it.

> I do not follow you.  I don't think this will bother you at all
> if you don't use it.

AFAICT its implementation has a cost even for those condition-case which
don't use the `debug' feature.


        Stefan

bug#3310: condition-case-no-debug

[Richard M Stallman]

    > I do not follow you.  I don't think this will bother you at all
    > if you don't use it.

    AFAICT its implementation has a cost even for those condition-case which
    don't use the `debug' feature.

Every feature has some cost, but I don't see why the cost of this feature
is enough to be worth worrying about.

bug#3310: condition-case-no-debug

[Lars Magne Ingebrigtsen]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

> Because condition-case-no-debug shows a simpler way to solve the same
> problem without any cost when the feature is not used (and also without
> touching any C code).
> We've been through this dicsussion.  All I wanted to say is to explain
> why we're in this situation.  I have no intention to get into the
> discussion again since we already agreed to disagree.

The function has a good doc string, so I think this bug report can be
closed.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/