Re: proposing patches (was: Colored kernel messages on serial console)

>> I find nothing wrong with proposing a patch, finding out that nobody
>> else finds it useful and so keeping it local.  [...]

Quite.

My apologies for reacting as knee-jerk negatively as I did.

> I actually thought the "sysctl for bytes to be prepended; sysctl for
> bytes to be postpended" proposal was rather clever.

I agree.  I can easily imagine it being used in unexpected ways, too,
which to my mind is one mark of a good choice of mechanism: it can be
used for what it was designed for but also finds other, unexpected,
uses.

Indeed, I can think of one other potential use already.  I once had an
unusual use case where I wanted the system to come up with an
almost-SLIP interface preconfigured...on the console serial port.  For
my use case, there was little enough console output that it didn't
interfere in practice, but the ability to bracket console output with
admin-chosen strings would make it feasible to keep console output from
interfering with the serial-line networking.

I'd actually be inclined to extend it so there are *two* start/end
string pairs: one for kernel output and one for `normal' userland
output to that tty, the latter specifically for the sake of the above
use case, in which there is a third data stream (the network traffic)
multiplexed onto the serial line.

/~\ The ASCII				  Mouse
\ / Ribbon Campaign
 X  Against HTML		[email protected]
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B