Re: git maintenance broken on FreeBSD

[Junio C Hamano <gitster@pobox.com>]

Todd Zullinger <tmz@pobox.com> writes:

> If a change like this is made, aren't we trading one group
> of broken users for another?  It would fix users of newer
> systems at the expense of those on older systems, I would
> suspect.

Thanks for raising this.  The description of POSIX "crontab"
command, cf.

https://pubs.opengroup.org/onlinepubs/9699919799/utilities/crontab.html 

talks about optional "file", but it is explicit that it has to be a
real file, i.e.

    file
        The pathname of a file that contains specifications, in the
        format defined in the INPUT FILES section, for crontab
        entries.

I would suspect that implementations may treat it as a sign to read
the standard input, but I do not think that is what the above
specifies.  For example, description of "file" argument of another
command, "diff", cf.

https://pubs.opengroup.org/onlinepubs/9699919799/utilities/diff.html

explicitly calls out that "-" stands for the standard input, i.e.

    diff [-c|-e|-f|-u|-C n|-U n] [-br] file1 file2

    file1, file2
        A pathname of a file to be compared. If either the file1 or
        file2 operand is '-', the standard input shall be used in
        its place.

So, it is fairly clear that "crontab" wants a real file.  Somebody's
POSIX compliant "crontab" can be fed "-", attempt to read from a
file with such a name, and legitimately fail.  And on such a system,
the proposed patch causes a regression.

> In the end, cronie adjusted it's behavior, which was similar
> to that of the newer vixie-cron, in 8b0241f (Partially
> revert the behavior of crontab command without arguments,
> 2021-03-17)¹.  It now behaves as required by POSIX if stdin
> is not a TTY.  That seems like a reasonable compromise and
> perhaps vixie-cron would be willing to do the same?

It indeed is a pragmatic solution to use isatty() as a hint.