Am Wed, 10 Feb 2021 00:38:15 -0500 schrieb Scott Kostyshak <skost...@lyx.org>:
> On Mon, Jan 25, 2021 at 02:02:09PM +0100, Kornel Benko wrote:
> > Problems with FindAdv and search with not ignored format
> >
> > What we have now is something 'neither fish nor fowl'.
> >
> > For instance for the search of characters with a specific size
> > we use strings/regexes wrapped with the size-spec. All is nice, until the
> > size is the
> > default in the document. In that case the wrapping has no effect (because
> > the text
> > created by latex output does not contain this info)
> >
> > Same is valid for language, color, family, series, shape, etc.
> > Also we don't have the options to exclude characters with specific features
> > from
> > search. (like search for words which are not part of a Latin sequence)
> >
> > I am not sure, if investing more into this (without creating a nex
> > export-type for
> > search (ala Docbook)) would get better sollutions.
> >
> > Comment are appreciated.
>
> I use Advanced Find about once a week and for those times it is just the
> right tool for the job. However, when I use it with certain combinations
> of regexes and settings, I do not not just use it without thinking about
> it and assuming that it does what I intended. I use it to do the
> find/replace or whatever I'm doing, and then I look at "git diff" of the
> document I modified to do a quick double-check, and then I do a manual
> search to see if I missed anything else. As long as I proceed with that
> caution, it's a helpful tool to me personally. And thinking about it
> more, when I do a regex find/replace in Vim I also always proceed with
> caution and do double-checking. I think this indicates that I really
> don't trust myself to get the regex right the first time.
This is excellent procedure.
> I don't know how to answer your questions since I don't know the code.
> But I feel it will be impossible to fix everything for it. Also, I think
> that this is a good example where having unit tests would be
> considerably better than the autotests.
Maybe, but still there has to be someone creating the tests.
> That is, without a unit test
> framework and an extensive set of unit tests, I would be hesitant to
> change anything.
Even with unit tests, you cannot be sure that the tests include all possible
cases.
But I agree.
> But that's just because the code looks scary to me.
> Also, I haven't tried the autotests in a while so maybe those work
> better than I remember.
I corrected some of the findadv tests, so they should work.
OTOH, they are way slower than the unit tests from Yuriy.
> I guess what I'm saying is I'm glad you spent a
> good chunk of time on it and fixed a bunch of issues all at once because
> I think whoever edits the code really needs to get a feeling for it.
> This part of the code does not seem like an area where one can casually
> drop in and fix individual bugs from time to time without unit tests
> that would check that nothing else is broken.
>
> In any case, thanks a lot for bravely tackling the code and improving
> it! I've been using it on master and it is more accurate and faster than
> before. My "trust" of it has definitely increased.
Thanks :)
> Scott
Kornel
pgpmsmAGmlBvV.pgp
Description: Digitale Signatur von OpenPGP
-- lyx-devel mailing list lyx-devel@lists.lyx.org http://lists.lyx.org/mailman/listinfo/lyx-devel
