bug#24426: 25.1.50; [critical] semantic-fetch-tags regression since 25-03-2016

bug#24426: 25.1.50; [critical] semantic-fetch-tags regression since 25-03-2016

[Oleh Krehel]

I have noticed a regression in the current master branch, possibly in
emacs-25 branch too.

When calling `semantic-fetch-tags`, it intermittently fails, even for
Elisp buffers for which it had a super-stable support over the years.

I noticed that semantic.el was not changed in the last years, so the
problem comes likely due to changes in EIOIO.

The error is thrown from `semantic-parse-region-default':

    (when (or (null semantic--parse-table) (eq semantic--parse-table t))
        ;; If there is no table, or it was set to t, then we are here by
        ;; some other mistake.  Do not throw an error deep in the parser.
        (error "No support found to parse buffer %S" (buffer-name)))

So I get e.g. signal(error ("No support found to parse buffer
\"hydra.el\"")), quite often, but not always.

The error definitely does not occur for an earlier build that I did on
25-03-2016. I'll try to bisect it more when I have some time.

I'd like to request putting this bug as blocking for Emacs-25. I heavily
rely on Semantic parsing the code in various languages and have some
packages on MELPA that rely on Semantic as well.

Oleh

bug#24426: 25.1.50; [critical] semantic-fetch-tags regression since 25-03-2016

[Dmitry Gutov]

Hi Oleh,

On 13.09.2016 11:09, Oleh Krehel wrote:
> possibly in
> emacs-25 branch too.

Could you double-check that first, please?

bug#24426: 25.1.50; [critical] semantic-fetch-tags regression since 25-03-2016

[Oleh Krehel]

Hi Dmitry,

Dmitry Gutov <dgutov@yandex.ru> writes:

> On 13.09.2016 11:09, Oleh Krehel wrote:
>> possibly in
>> emacs-25 branch too.
>
> Could you double-check that first, please?

Just double-checked. Can no longer reproduce the bug on master (and
emacs-25). The bug was related to a corrupted semanticdb I guess, fixed
by cleaning that folder.

I'll reply here if the bug resurfaces, but it's gone for now.