[bug#43194] [PATCH] gnu: publicly define freedink-engine and freedink-data

[Jesse Gibbons <jgibbons2357@gmail.com>]

Thank you for reviewing.

On 9/7/20 7:46 AM, Ludovic Courtès wrote:
> Hi,
>
> Jesse Gibbons <jgibbons2357@gmail.com> skribis:
>
>> The attached patch publicly defines freedink-engine and
>> freedink-data. This resolves many of the issues described in
>> #43061. This patch, combined with patch #43193(sent earlier today),
>> can close #43061.
> Now I’m confused: how does it help to make freedink-{engine,data}
> public?
Other than making guix more consistent in publicly defining game data 
packages (0ad-data and megaglest-data are public, and I like that -- I 
could write a good article about why, which I think would be a worthy 
entry in the guix blog, especially after #43193 is applied), there are 4 
reasons for this change:

-> freedink-dfarc has problems locating the editor, installed in 
freedink-engine. I guess we could also fix this by making 
freedink-engine an input to freedink-dfarc and splicing a reference to 
it into the default configuration?

-> Unless freedink-data is public, `guix build --source freedink-data` 
fails, and `guix build --sources=all freedink` does not build a source 
for freedink-data. I think future users who want to alter the freedink 
data would appreciate the ability to use guix to get the data. Also, 
it's pointless to use the editor on the installed freedink-data because 
it's read-only when it's installed.

-> Back when I was fixing freedink, I found it difficult to debug 
without freedink-engine being public, because freedink does nothing with 
the freedink-engine source.

-> Freedink-engine installs desktop files to launch freedink without 
freedink-dfarc or the console. This is actually a new issue I will 
address in an updated patch: the desktop files fail because the data 
location is not hard-coded. I think the freedink desktop file can be 
patched if freedink-data is an input, but, like I said above, it's 
pointless to use dinkedit on a read-only directory, so I intend to 
remove it.

>
>> >From 583215aced9b557d6f4e54b290e788d33880c03c Mon Sep 17 00:00:00 2001
>> From: Jesse Gibbons <jgibbons2357+guix@gmail.com>
>> Date: Wed, 26 Aug 2020 21:38:24 -0600
>> Subject: [PATCH v1 1/1] gnu: publicly define freedink-engine and freedink-data
>>
>> * gnu/packages/games.scm: (freedink-engine): make public
>> (freedink-data): make public
> [...]
>
>>   (define-public freedink
>>     ;; This is a wrapper that tells the engine where to find the data.
>> -  (package (inherit freedink-engine)
>> +  (package ;(inherit freedink-engine)
> Is it intended?  Looks like inheriting avoids duplicating fields, no?

Oops! I did not intend to leave (inherit freedink-engine) in a comment. 
I initially commented it out because freedink does nothing with the 
source anyway, and I wanted to see what would happen if I removed the 
inheritance. I guess I forgot to remove the semicolon and other additions.

As noted above, it is easiest to use freedink-dfarc to launch the 
editor, but freedink-dfarc must be told what editor to use, and it is 
easier to identify it if the editor is installed in a profile (or 
included as an input). Also, freedink-engine includes (broken) desktop 
files. Since freedink just installs a wrapper script around the engine, 
and does not include the editor or any desktop files, perhaps it would 
be better to put the wrapper script in freedink-engine (thus fixing the 
desktop file), completely remove the freedink package, and rename 
"freedink-engine" to just "freedink"? But freedink-dfarc would still 
need to be able to launch freedink without pointing to any read-only 
data if a user wants to test the edited freedink data.

>
> Thanks,
> Ludo’.
-Jesse