Re: 71ca97da9d overrides user's kernel config

Re: 71ca97da9d overrides user's kernel config

[Linus Torvalds]

On Fri, Jun 1, 2012 at 8:30 AM, Vladimir 'φ-coder/phcoder' Serbinenko
<phcoder@gmail.com> wrote:
>
> Sorry, it wasn't intentional. It was just a stub to compile it without
> recompiling whole kernel. Should I supply the patch to remove it?

Yes, but I wonder if they should *also* be renamed, actually.

Those names are disgusting and unreadable, because the "MAC" part
mixes with the actual name of the codepage. I really have a hard time
reading something like MACROMANIAN, because to me it looks like it
starts with "MACRO". It's just a bad name.

Also, the CODEPAGE part adds questionable value. I wonder if we should
(a) remove the CODEPAGE part, (b) add an underscore between "MAC" and
the name, and for the microsoft numeric codepages, do the same just to
be consistent. IOW, use config names like NLS_DOS_1250 and
NLS_MAC_ROMANIAN?

(Yes, yes, I realize that people think that 1250 is a "windows"
codepage, not a DOS codepage. Use WIN_1250 if you want to, but I think
it's all really the same.)

I dunno. But start with the Mac ones first, since they need fixing
anyway. Leave the old ones alone for now.

Andrew, who would be in charge of "fs/nls"? Almost nobody has really
touched that area in several years now, and it doesn't have any
maintainership. Added the few poor sods that touched it last. "Tag,
guys, you're it!"

                 Linus

Re: 71ca97da9d overrides user's kernel config

[Vladimir 'φ-coder/phcoder' Serbinenko]

[-- Attachment #1: Type: text/plain, Size: 1478 bytes --]


> Yes, but I wonder if they should *also* be renamed, actually.
> 
> Those names are disgusting and unreadable, because the "MAC" part
> mixes with the actual name of the codepage. I really have a hard time
> reading something like MACROMANIAN, because to me it looks like it
> starts with "MACRO". It's just a bad name.
> 

I agree. It was something I've seen used as charset name but quick
websearch shows better alternatives like "Mac Romanian", "Mac OS
Romanian". Should I rename the charsets themselves as well?

> Also, the CODEPAGE part adds questionable value. I wonder if we should
> (a) remove the CODEPAGE part, (b) add an underscore between "MAC" and
> the name, and for the microsoft numeric codepages, do the same just to
> be consistent. IOW, use config names like NLS_DOS_1250 and
> NLS_MAC_ROMANIAN?

I'm ok with changing my newly-added charsets but I think that there is
no reason to rename already existing ones. It would just add a bunch of
new questions at reconfigure not corresponding to any real
functionality. Also, I believe, that the whole reason of NLS codepage
tables to exist is the ability to read old volumes and marginally the
ability to interchange data with old OS in a best available way. The
least it used for any purpose outside speaking with old OS or touching
old volumes, the better.
I'll prepare 2 patches: cleanup, config rename and codepage rename.
-- 
Regards
Vladimir 'φ-coder/phcoder' Serbinenko


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 294 bytes --]

Re: 71ca97da9d overrides user's kernel config

[Andrew Morton]

On Fri, 1 Jun 2012 10:53:20 -0700
Linus Torvalds <torvalds@linux-foundation.org> wrote:

> Andrew, who would be in charge of "fs/nls"? Almost nobody has really
> touched that area in several years now, and it doesn't have any
> maintainership.

I cc'ed various people on that patch who have had fingers in there in
recent years but no, it doesn't have a maintainer.