uniq truncates lines > 2048 bytes

[Tim Robbins]

On Tue, Jan 25, 2005 at 11:51:51AM -0600, Scot Hetzel wrote:
> I noticed that if a file has lines > 2048 bytes, uniq will truncate
> the line to LINE_MAX (2048 bytes). An easy way to test this is to do
> the following:
> 
> cd /usr/ports/accessibility/gnomemag
> make fetch-list > test.list
> make fetch-list >> test.list
> uniq test.list > test2.list
> 
> test2.list should be half the size of test.list, but it is 2048 bytes.
> 
> I have come up with a patch to uniq that fixes this problem.
> 
> http://www.freebsd.org/cgi/query-pr.cgi?pr=76578

This looks good except for failure to check for realloc() returning NULL
and a few minor style problems. It may be possible to use fgetwln()
to read lines instead of getwc() + realloc() etc., but this function is
new and peculiar to FreeBSD.

I was planning on going through all text-processing utilities in the base
system some time and either fixing line length problems or documenting them,
similar to what I did with multibyte character support. I may make a start
at that today.


Tim