fuzzy-matching in quickopen...

[Waqar Ahmed]

> Are you sure this is a better behaviour ?

It doesn't matter. With a single letter one cannot really get very far and
can't really do much scoring. So I wouldn't waste time on thinking about
the single typed letter case.

> which ? I thought that's what quickopen is for, really. At least it was
until
a year ago.

No, that is not what it is for. If it were, it wouldn't even list other
files at all. Quickopen is for opening files in your project + any already
open files (which may or may not be in the project)

And quickopen is two years old by now. It was already there when you were
making those search/replace patches.

Tab switcher / documents plugin deal with already open files.

You can try increasing the score of already open files. Maybe it will make
it better, maybe not. Right now only one point is added, which is probably
too low. Maybe there is a better way to add this score so that files that
are a good match + open are ranked better.



On Sat, Sep 17, 2022, 1:02 AM Alexander Neundorf <neundorf at kde.org> wrote:

> Hi,
>
> On Freitag, 16. September 2022 16:25:15 CEST Waqar Ahmed wrote:
> > 1. After filtering, why should open files always end up at top? This
> > is not an "already open files quickopen". If the filter matches
> > something else better that will end up at top. There might be cases
> > where an openfile is a better match, perhaps there is a way to improve
> > that without biasing 100% in favour of open files.
>
> From what I remember this started as a "already open files quickopen", and
> just
> later on got extended to also show files which are not open yet. This was
> great.
> You are asking why the files which are already open should be at the top:
> my typical use case (working as a software developer for many years now
> already, basically always using kate) is that I open a bunch of
> interesting
> files, and then switch between those files very often. From time to time a
> new
> file joins the round, but this is much less often.
> Currently I'm working on a project where a source tree consists of several
> thousand files. The fuzzy match very often finds something different than
> what I
> wanted, and I know I typed the 3 or 4 characters correctly.
> I know in these cases that the file I want to go to is among the probably
> 10,
> or maybe 20 files I looked at before, so the quickopen saved me from
> searching
> that file manually in the top of the list. Now I start typing and
> seemingly
> random files appear at the top, and I have to scroll down to find the file.
>
> Maybe sorting by the score is (at least for the open files) not such a
> good
> idea. I would prefer if the order of the files would actually stay the
> same,
> and by typing it would just filter files out from that.
>
> > 2. Match in sequence will take precedence once you have typed 4 or
> > more letters. With 3 or less letters, we can't be sure if that is a
> > sequence or an abbreviation. e.g., ftv will prefer "FilesTreeView"
> > over "abcftv.js"
> >
> > 3. That is correct and working as expected.
>
> Ok. Are you sure this is a better behaviour ?
>
> > Quickopen is not meant to filter already open files. For that, you
> > have other plugins that can do the job.
>
> which ? I thought that's what quickopen is for, really. At least it was
> until
> a year ago.
>
> Sorry, probably my mail sounds a bit negative, but I really just want to
> make
> sure kate is working well :-)
> For the conrete example, I think  should not post the file names from my
> day
> job.
>
> Thanks
> Alex
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kwrite-devel/attachments/20220917/7e577ecb/attachment.htm>