Re: [AUCTeX-devel] Add `unit' type for the parser in siunitx.el

[Tassilo Horn]

Mosè Giordano <address@hidden> writes:

Hi Mosè,

>> That doesn't sound like the right solution.  Too much fuzz for such
>> an uncommon use-case, IMHO.
>>
>> Maybe the style hooks should be run inside a `condition-case', and in
>> case of an error we simply run `TeX-normal-mode'?
>
> Yes, I know it's not a common case, indeed I didn't notice this in my
> first tests.  But when it comes it can be annoying for a user that
> don't know what's happened since buffer is not even switched to the
> LaTeX source.

Sure it's annoying and should be improved.

> The idea of using `TeX-auto-store-pre-string' came to me when thinking
> about a way to store in parsed file the options given to packages at
> load time.

I don't see how storing package options is related to your
`TeX-auto-store-pre-string'.

> Do you have alternative suggestions for this?

IMO, it would make sense to simply discard all auto-parsed information
in case loading the auto/foo.el signals an error.

> If they are OK, can my changes without `TeX-auto-store-pre-string' be
> committed?  I'm sending the last patch for `siunitx.el' without those
> lines.

Yes, I did so.  I made a small change in that LaTeX-siunitx-regexp is
just the regex now [not a list (regex 1 LaTeX-auto-siunitx-unit)].  The
latter is created at the call to `TeX-auto-add-regexp' in the lambda
being the style hook for siunitx.

And I have one more question: The new type for the parser is for siunitx
units.  However, LaTeX-siunitx-regexp says it matches unit, power,
prefix, and qualifier definitions.  Is that all the same, e.g., may I
provide a qualifier where a power was expected?  Or should there be a
individual parser type for each siunitx thingy?

Bye,
Tassilo