Subject: Re: bin/1697: vi bug (mmap() of remote file is not consistent)
To: Mike Long <mike.long@analog.com>
From: Jason Downs <downsj@teeny.org>
List: netbsd-bugs
Date: 10/30/1995 20:52:07
In message <9510301833.AA14333@cthulhu>,
	Mike Long writes:
>der Mouse <mouse@Collatz.McRCIM.McGill.EDU> wrote:
>
>>> >Synopsis:       vi uses mmap() instead of read(); can corrupt remote files
>>
>>This is a bug in the VM subsystem, exhibited by mmap, not a bug in vi.
>>"Fixing" it by changing vi would be like "fixing" a compiler that
>>botched typedef declarations by using #defines instead of typedefs.
>
>Your analogy oversimplifies the problem.  VM/buffer cache integration
>is a major architectural change to the kernel; I don't expect that it
>will happen anytime soon.
>
>A precedent for the desired change to vi may be found in /bin/cp; the
>mmap() code was commented out so that cp now uses read() exclusively.
>vi should be fixed similarly, so that use of mmap() may be easily
>restored when the buffer cache integration does occur.

It is typical of NetBSD to postpone making something work correctly
(such as vi), in favor of some lofty goal that will take some time to
reach, if ever.

One would much rather have a working vi, now, then wait another year.

--
Jason Downs
downsj@teeny.org  --> teeny.org: Free Software for a Free Internet <--