Re: [PATCH] Feature: custom guitool commands can now have custom keyboard shortcuts

From: Harish Karumuthil <harish2704@gmail.com>
Date: Oct 5, 2019, 20:16 UTC
Message-ID: <149a83fd40b71896b134b16c2b499ff472c6234e.camel@gmail.com>
In-Reply-To: <20191004120107.kpskplwhflnsamwu@yadavpratyush.com>

Hi All,

From https://www.kernel.org/doc/html/v4.10/process/email-clients.html, I understood that, my current email client ( that is gmail web ) is not good for submitting patches. So I was tying to setup a mail client which is compatible with `git send-mail`. But I was not able to get a satisfactory result in that.

For now, I followed the instruction of Johannes Schindelin and submitted a pull request . Please see https://github.com/gitgitgadget/git/pull/376

--------- @ Pratyush: Regarding your comments,

> A pretty nice way of doing it. But I would _really_ like it if there was
> an option in the "create tool" dialog to specify the shortcut. People of
> a gui tool shouldn't have to mess around with config files as much as
> possible.

I agree with this, But that may require some more profficiency in TCL/TK programming which I don't have. This is the first time I am looking into a TCL/TK source code. Any way I will try to integrate the gui gradually in feature. But unfortunatly, I may not be able to do that now.

Show 8 quoted lines

> David has advocated inter-operability between git-gui and git-cola.
> While I personally don't know how many people actually use both the
> tools at the same time, it doesn't sound like a bad idea either.
>
> So, sharing shortcuts with git-cola would be nice. Of course, it would
> then mean that we would have to parse the config parameter before
> feeding them to `bind`. I don't suppose that should be something too
> complicated to do, but I admit I haven't looked too deeply into it.

IMHO, Using a uniform shortcut-key code/foramat for both application can be considered as nice feature. But, whether we should share common shortcut-scheme with both application is a different question. Currently, both apps don't have a common shortcut-scheme. So in this situation, only sharing custom-tool's shortcut-scheme with both applications doesn't look like a good idea to me

Show 6 quoted lines

> Are you sure that is a good idea? I think we should at least make sure
> we are not binding some illegal sequence, and if we are, we should warn
> the user about it. And a much more important case would be when a user
> over-writes a pre-existing shortcut for other commands like "commit",
> "reset", etc. In that case, the menu entires of those commands would
> still be labelled with the shortcut, but it won't actually work.

I agree with you. It is an important point. After reading this, I checked current status of these issues. What I found is given below.

1. When user provides an invalid sequence for the shortcut, it will cuase the
entire gitgui application to crash at the startup

2. When user tries to overwrite existing shortcut, it will not have any
effect. Because, built in shortcuts will overwrite user provided one. But
still, wrong menu accelerator label will persist for custom tools

Since #1 is a serious issue, I tried to find out the function which does the keycode validation, but I haven't succeded till now. ( I found the C function name which is "TkStringToKeysym" from TK source, but I couldn't find its TCL binding ). It will be helpful if any one can help me on this.