* Herbert Xu said: > > No, the issue is what behavior Debian systems (and other operating > > systems based on the Linux kernel) expect "echo" to have. Judging > > from the comments from the LSB list (and I think it's safe to assume > > They only care about the shell, not shell scripts. I presume they are going > to stick with what POSIX.2 says, which is that no portable script should use > options. If they are actually going to mandate -e/-n, i.e., break POSIX.2, > then it would be a completely different matter. You still seem to put the equal sign between what the shell supports and the scripts interpreted by the shell. The shell is *your* product, the script is somebody *else's* - both of you make decisions based on a) standard, b) common practice, d) environment the product is intended for. a) is POSIX which allows for -n/-e, b) requires the two switches to be present for a huge base of scripts to work, d) Debian/GNU Linux supports the switches in each and every shell in the distribution - you mustn't break the status quo, and the environment for a sake of strict compliance with a required part of the standard which otherwise allows the variations to exist. > > Now, if ash is not going to be compliant with LSB, and behave > > differently from other Bourne-compatible shells, then we should not > > use ash as an alternative for the standard Bourne-compatible shell > > (/bin/sh). OTOH, since it's pretty obviously trivial to make ash an > > acceptable alternative (by allowing and acting on the switches), there > > is no good reason not to do so. > > Our policy currently requires #!/bin/sh scripts to be POSIX compliant. If > you don't like it, then get it changed. But when that is done, it will be no > longer possible use anything other than bash as /bin/sh. But scripts using -n *are* POSIX compliant *if* they account for the possibility that the argument might not be implemented in some shell. marek
Attachment:
pgpNuAEcq_O76.pgp
Description: PGP signature