Re: 2 minor items to GRUB 0.9x

[Christoph Plattner]

Hello,
thanks for the answer.

It is also important for me to check, how important which feature is.
But here some words, why I introduced those points:

XON/XOFF: From Terminal to GRUB (GRUB incomming side) I often
encountered the problem on the command line (and menu), that using the
mouse (cut&paste) or using some tools like "expect", GRUB looses
characters, when it shifts the line, etc...

This very small patch (only for the incomming side) could help.
For the outgoing side I never saw a problem up to now, but for complete-
ness it should be added too, but with low priority.


"--diskless-only": This is a very small patch, but very important
in my oppinion. You (Mr. Okuji) designed a build process to build
up GRUB for local boot and GRUB build up for diskless boot
simultaniously.

But the current implementation of GRUB - when using the preset menu -
breaks the build process, as now two different types of preset menu
are necessary, which is not possible in one configuration.

So I see 2 primitive solutions, and those are NO "ad hoc" ideas,
I often discussed solutions here.

*1* The current code in stage2.c uses `bootp\n' as preset menu in
the diskless mode, if no preset menu is defined.
To solve the problem described above, we should also add `bootp\n'
to the end of the preset menu, and not include the possible `bootp'
in the preset menu.

*2* Give the `bootp' the possibility, to be only executed (valid),
if the GRUB is really in the diskless mode (I mean, that it is
the diskless build #ifdef SUPPORT_DISKLESS...#endif). This can be
done with the option "--diskless-only" (or whatever name make
sense here).

So my conclusion:

XON/XOFF: The default serial console on UNIX machines use XON/XOFF,
and GRUB should also follow this standard to have "complete serial
console support". For the practical use, the incomming flow control
may be important, the outgoing is unimportant.

DISKLESS/preset menu: I think this patch is very trival, and I think
it is important not to break your good design idea of parallel build
of local and diskless GRUB.

With friendly regards
Christoph P.



Yoshinori K. Okuji wrote:
> Basically, it depends on how crucial it is whether a given feature is
> integrated or not. Any code change could make software unstable, as
> you know, so I'll install a new feature, only if you can't live with
> GRUB without such a feature. Non-critical features are never added.
>
> If you believe that something is necessary, describe why it is so
> important. It would be a good idea to show the very problem you're
> faced with.
>
> According to that policy, it is not clear why XON/XOFF support is
> important, so I can't add it.
>
> As for --diskless-only, I suspect that it is too "ad hoc". And, I'm
> wondering if it is so important to share the same configuration. How
> difficult do you maintain two configurations?
>
> Thanks,
> Okuji
>
>
> _______________________________________________
> Bug-grub mailing list
> address@hidden
> http://mail.gnu.org/mailman/listinfo/bug-grub



--
-------------------------------------------------------
private:        address@hidden
company:        address@hidden