[illumos-Developer] Requesting code review for opensource /usr/xpg4/bin/more replacement with "less" ...

[I. Szczesniak]

On Tue, Jan 18, 2011 at 4:07 PM, Garrett D'Amore <garrett at damore.org> wrote:
> On Tue, 2011-01-18 at 10:35 +0100, Joerg Schilling wrote:
>> "Garrett D'Amore" <garrett at damore.org> wrote:
>>
>> > On Mon, 2011-01-17 at 11:54 -0800, Adam Leventhal wrote:
>> > > Hey Roland,
>> > >
>> > > Apologies if I missed this discussion, but why is this something that
>> > > should be in illumos? First less is not the same as more; second, why
>> > > wouldn't the downstream distros just include less if that's what they
>> > > wanted to do?
>> >
>> > Actually, it *is* the same as more, when called as "more".  In fact, the
>> > author of the original "more" utility on some platforms is the same as
>> > the author of "less".
>>
>> I cm not willing to believe this.
>>
>> more has a careful attempt to deal with the terminal that even works in case
>> that there is no TERM set up, while less has an agressive attempt to deal with
>> the terminal (to be specific, it deals with the terminal as if it was an
>> editor) that is known to often fail.
>>
>> A quick check with calling /usr/bin/less as more (more /etc/passwd) shows a
>> flicker and then it looks as if nothing happened.
>>
>> It would be nice if things were so simple that you really can use less to
>> replace /usr/xpg4/bin/more.
>>
>> Jörg
>>
>
> I think the problem you are seeing is a bug in xterm or TERMCAP, not in
> less.
>
> Specifically, less behaves as more -e by default (which might be argued
> to be a bug) when called as more.
>
> Given a short file (say /etc/motd), the output from less is not
> displayed on the terminal.  But truss shows that it did indeed write()
> to the tty.
>
> Notably, XPG4 more -e suffers from the same bug.
>
> To reproduce, do
>
> % printf "abc\n" > /tmp/a
> % /usr/xpg4/bin/more -e /tmp/a
>
> Surprisingly, piping to /bin/cat resolves the problem.
>
> Also, setting TERM=dumb resolves the problem.
>
> I'm not sure yet if the bug occurs on the console.
>
> I'd be willing to accept that fixing this problem (which might just be
> something stupid in the TERMCAP) is a pre-requisite for getting this
> integrated.

Why? Oracle's /usr/xpg4/bin/more exhibits the same problem. Neither
does it pass all tests of VSC. Why can't you integrate this patch and
then work on improvements in incremental steps?

Irek