Re: [PATCH 0/2] mergetools: vimdiff3: fix regression

From: Fernando Ramos <greenfoo@u92.eu>
Date: Aug 6, 2022, 21:23 UTC
Message-ID: <Yu7byRmn4VtSpyWd@zacax395.localdomain>
In-Reply-To: <CAMP44s3-RG5k4ZkhAFG_9JtbxcyDhkUmeBh0jCH9+Xwyumyu9w@mail.gmail.com>

On 22/08/06 02:17PM, Felipe Contreras wrote:

Show 8 quoted lines

> 
> I don't know why anyone would want to do that, but the code interprets
> that as the user wanting '1b', which is completely ignored.
> 
> If we are not going to care about these cases, we can just remove all this code:
> 
> ...
>

Ah! I see now. You are completely right: it wouldn't make sense for anyone to specify "layout=LOCAL" (or REMOTE or BASE), but if he did *it wouldn't work* (only works with "layout=MERGED").

That should be fixed. I'll update the patch with a new version to generate this command string:

     echo | silent 4b | set hidden | let tmp=bufnr('%') | silent bufdo diffthis | exe 'buffer '.tmp
                                     ^^^^^^^^^^^^^^^^^^                           ^^^^^^^^^^^^^^^^^
                                     NEW                                          NEW

Notes:

  - This is "easier" than moving "silent 4b" to the end, due to the way the
    code is structured.

  - I agree that this is absurdly complex for what we want to achieve with
    "vimdiff3" but let's put it this way: now everything can be achieved with
    the "layout" configuration option, even "useless" things such as setting it
    to "LOCAL".

> I understand the need if you want a complex layout, like
> "MERGED+LOCAL,BASE,REMOTE", that's very nice, but if you just want
> "MERGED", most of the code does nothing,

With the fix above that shouldn't be a problem anymore: even if someone specifies "LOCAL" it will work, in an absurd way, but it will work :)

> the extra -c "tabfirst" isn't needed either.

Good catch. I'm also removing it.