Re: [PATCH 1/1] Kconfig: Added compare capabilities for mconf

[Franco Martelli <martellif67@gmail.com>]

On 11/11/25 at 15:48, Nicolas Schier wrote:
> Hi Franco,
> 
> thanks for your contribution.  Some notes below:

Hi Nicolas,

thank you very much for your answer.

> 
> On Thu, Oct 30, 2025 at 03:13:42PM +0100, Franco Martelli wrote:
>> This patch allows to compare the tristate value of two .config files
>> inline, symbol by symbol, in mconf (make menuconfig) only. It can be
>> really useful when customizing the .config file.
>>
>> When a new stable release of a distribution is available, in the Linux
>> kernel many things has changed, this patch comes in help in this
>> situations, it allows to easily compare the customized .config file
>> built for the previous version of the distribution, with the one
>> provided by the new software distribution, used as reference.
>> Generally it can be helpful when the differences between .config files
>> are huge, because you can navigate across them in the menu window.
>>
>> The patch has a limitation, it only show the tristate value (Y/N/M) not
>> differences for symbol with a value (string), but it also marks with a
>> "#" the symbol missing in the loaded .config, meaning therefore the
>> symbol as a new feature.
> 
> I must confess, that I am not convinced yet, that this will be helpful
> for general purposes.  Yes, large changes in .config files are not easy
> to parse, but I do not see how your solution is easier to handle than
> the use of scripts/diffconfig, other than not showing removed config
> options.

the strong of my patch is interaction; scripts/diffconfig merely show
you a list of differences that you'll have to handle instead, having
differences reported near every symbol's item in the mconf window let
you to interact with the differences inline, on the fly.

> 
>>
>> o I've also added support for the more canonical <F1> key to show inline
>>    help window, in addition to <?> and <h> keys.
> 
> Please split-off logical these independent changes into separate
> patches (Add support for 'F1' hot-key).

OK, sorry!

> 
> Using simple letters (here 'h') as hot-keys is not a good idea as those
> are already used for jumping through the currently shown menu item
> (cp. highlighted letters in menu config screens).

the 'h' hot-key was already in the source code used to open the 'help'
window, it wasn't a my change.
Using simple letters to call functionality was strongly advised to me by
Randy Dunlap because some desktop environments intercept function keys
by default, using letters provides a workaround for this.

> 
>> o By pressing the <F2> or <D> key you will get a view with all the
>>    symbol differences listed.
>>
>> o By pressing the <F4> or <F> key you'll save to a file the differences.
> 
> While the list might be helpful for comparing two config files, a large
> list within an ncurses window does not look like a good thing to me.

True but the list will (hopefully) shrink during the configuration
process and the list now (version 4) is sorted alphabetically.

> 
> When I save the list to a file (F4), I'd rather choose the output of
> diffconfig for evaluation -- as it is more complete and has a layout
> that looks more straight-forward and clear to me.

OK, it seems easy to implement.

>> o Added help text that it describes the new features.
>>
>> This software is under GPL-2.0 license.
> 
> Please note that the whole Linux kernel source and its build system is
> GPL-2.0 per se, cp. COPYING and the SPDX header line in each file.
> 
>>
>> Changelog:
>> ----------
>>
>> scripts/kconfig/confdata.c:
>> ---------------------------
>>
>> Silenced a warning emitted when parsing the .config file to compare.
> [...]
> 
> Please put any patch changelog below the triple-dash line ('---').

OK sorry, I'm new to the kernel process patching.

> 
> I have not looked at the code, yet, as I have doubts regarding the
> usefulness (see above).
> 
> Iff nobody mentions strong interest in these changes, please re-evaluate
> if it makes sense for you to invest into a v5.

I'll wait a couple of week, if in the meanwhile nobody appreciate my
patch and therefore it will be rejected I'll made it available on my
Github profile for the Linux 6.12.y kernel series.

> Kind regards,
> Nicolas

Thanks again, kind regards.
-- 
Franco Martelli