>Jonathan Stone <jonathan@DSG.Stanford.EDU> writes:
>> Last: my impression is that most of the actual problems with sysinst
>> are due to the x86 MBR handling.  Lest some people forget,
>> most of the feedback on 1.3/pmax was along the lines of:


>Strongly disagree.

On the design or on the feedback?  If its about the feedback, then you
really need to check the archives.

On the design: point taken. s/actual/chronic/.


>Go look again through some of the remaining sysinst PRs, and the ones
>I recently fixed.  There are plenty of things which have nothing to do
>with i386 MBR handling.

Can't, or I would have checked before posting.  Our websever is
`having problems', remember?


>There are scores of problems in the disk partitioning code (non-MBR; I
>_never_ do special MBR setups but i've run into N problems with the
>partitioning code), in the display/menu code, in the install media
>selection code.  Some of them may not be bugs, i.e. the code as
>designed and written is functional, but badly misdesigned.

PR them by the score, then.  I agree about the disklabelling code:
thats one of the very first things I noticed.

But (as Ive said for ages) one of the big problems with sysinst is
that there's no-one in charge of it. So it accretes random features,
without thorough testing or UI consideration.  _That_ is the biggest
problem, and its not one that a rewrite is going to fix in the long term.