Re: Compiled code in Emacs-26 will fail in Emacs-25 if use pcase [was: Add new bytecode op `switch' for implementing branch tables.]

[raman]

Tino Calancha <address@hidden> writes:

Likely related, I'm also noticing
 that code that compiles and run correctly against 25.1 and that used to
 compile and run error-free on Master just started breaking,
I'm still working on creating a small repro case.


But what I'm noticing:

With code compiled with lexical scoping,  defsubst forms appear to
behave differently when compiled vs evaled --- eval: no error, compile
and load == error.
> Sorry if this is obvious, then just ignore.
>
> I have a custom lib using `pcase' somewhere.  That lib is
> compatible with Emacs25.
> Usually i compile this lib against the master branch.  Then, i load
> that very same resulting .elc in Emacs-25 and runs same as in Emacs-26.
> (Maybe this is just wrong and i must keep separated .elc)
>
> After the new bytecode those parts of the lib using pcase fail.
> For instance, put the following in a file:
> ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
> (defun test-case ()
>   (let ((val "foo"))
>     (pcase val
>       ("foo" "foo")
>       ("bar" "bar")
>       (_ (error "Neither 'foo' nor 'bar'")))))
> ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
> I) Compile under master branch.
> II) Load the resultant .elc with Emacs-25
> II) M-: (test-case)
> ;; Signal error: Invalid byte opcode: op=183, ptr=4
>
> So, in conclusion:
> From now on, we need to keep two separated .elc in such custom libs,
> right?
>

--