Why is ,. (comma-dot) treated specially?

Why is ,. (comma-dot) treated specially?

[Artur Malabarba]

I ran into a problem the other day where writing
    `(message ,.some-var)
unexpectedly split the .some-var symbol without evaluating anything:
    (message (\,\. some-var))

Meanwhile, adding a space after the comma:
    `(message , .some-var)
yielded the expected result:
    (message [value of .some-var])

Upon inspecting `lread.c', I see that indeed ",." is treated specially
(look for references to Qcomma_dot), similarly to how ",@" is special.
On the other hand, I can find no mention of this "comma-dot" anywhere
inside backquote.el.

Is this a feature that was planned but is not yet finished? Why is the
dot handled specially after a comma if we don't do anything with it?



Cheers

Re: Why is ,. (comma-dot) treated specially?

[Stefan Monnier]

> Is this a feature that was planned but is not yet finished? Why is the
> dot handled specially after a comma if we don't do anything with it?

Using special characters (like dot) in symbols is risky.  Always has
been always will be.

> I ran into a problem the other day where writing
>     `(message ,.some-var)
> unexpectedly split the .some-var symbol without evaluating anything:
>     (message (\,\. some-var))

I'm not surprised it doesn't do what you expected, but I'm admittedly
surprised of the actual behavior.  In any case, this looks like
a plain bug.


        Stefan