On Thursday 04 June 2009 01:28:01 Giacomo Catenazzi wrote: > Raphael Geissert wrote: > I think we could forbid "echo -n" on maintainer and other scripts, but > allow it on init.d scripts.
Like Julien said, I think making different recommendations is a bad idea. > Rationale: > - portability is more important on other scripts, and it is also not so > frequent to use echo -n No, not really. In a large project the interest is to be as portable as possible, diverging as less as possible to support different ecosystems. And that applies to all kind of code. > > Yes, printf. > > It is not an alternative: > - It is ugly You say "it is ugly", I say "it is pretty", and none of us can be proved right or wrong because it is a subjective statement. Rephrasing or otherwise omitting that statement would be better for the sake of the discussion. > - it is not on root partition Only when not implemented as a shell built-in (which is very very rare, and I think of all the shell interpreters in Debian only posh doesn't implement it because it is not required by policy.) However, if it is a real concern then we could request printf to be installed in /bin. > Most of people understand "echo -n", but I don't know if all people > understand fully printf. Mixing echo and printf is ugly. printf exists as-is or has an equivalent in most programming languages, so I doubt a more or less formed programmer won't know how printf works. OTOH, echo -n is just a bashism, not even an XSIsm[1], and the option name doesn't tell much about what it does. Not to mention that this discussion is not only about *discouraging* (not forbidding!) the -n option, but when passing a string with backlashes to echo. > Anyway the standard are helper tools, not tools to make difficult the live > of user and maintainer. 'echo' is a portability nightmare if not used properly, I don't know how you could say it helps. > See for example the recent discussion about ext3, > see the old discussion about tar (tar is not in POSIX, people should use > 'pax') I think that's irrelevant, and for the sake of this discussion I would like if you could stop giving examples of how "we are not POSIX." The reasons for that, I think, are pretty obvious. > > > BTW I think this is incorrect: > $ POSIXLY_CORRECT=1 bash -c "echo -n line one; echo line two"; > > line oneline two Policy doesn't make any exeption, so the behaviour follows what policy mandates. It is just another example of why we can't rely on echo. [1] " On XSI-conformant systems, if the first operand is -n, it shall be treated as a string, not an option." Regards, -- Raphael Geissert - Debian Maintainer www.debian.org - get.debian.net -- To UNSUBSCRIBE, email to debian-policy-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
