Man vs. standards
in [Shell]
|
Prev: stratifying values in a file ...
Next: find errors "paths must precede expression: %d" ..., "missing argument to `-exec'"
From: Ivan Shmakov on 4 Feb 2010 16:20 >>>>> "S" == Seebs <usenet-nospam(a)seebs.net> writes: >>>>> "IS" == Ivan Shmakov <ivan(a)main.uusia.org> wrote: [...] S> The way to show that something is not standard would be to show that S> the standard doesn't specify it, not to appeal to whether a S> particular system has it. Many systems lack many features which are S> clearly standardized. IS> I didn't say that they're not standard. I've just noted that IS> they're out of necessity on these particular systems. S> I have apparently lost the thread. I asserted that they were a S> standard tool for a task. You showed examples of systems which S> lacked that tool. S> I do not understand what you mean by "out of necessity". Actually, I was trying to tell just the same, “many systems lack many features which are clearly standardized.” But now, I cannot understand how the “standardness” of the tool can be of any help given that I cannot expect the system I work with to conform to the standard? The deviance to the other parts of POSIX manifests itself early: one cannot expect a reasonable behaviour of almost any software tool built and running on top of a system having a flaw in the open () system call. On the other hand, section 2 Man pages seem somewhat extraneous for me to install on these systems given that I haven't received even a single request (IIRC) for them to be installed. Being standard and being useful are two different things. There may be non-standard systems which are useful, yet there may be standard ones which aren't. There's just too little reason for a screw to be a standard nail. -- FSF associate member #7257
From: Seebs on 4 Feb 2010 16:28 On 2010-02-04, Ivan Shmakov <ivan(a)main.uusia.org> wrote: > But now, I cannot understand how the ?standardness? of the tool > can be of any help given that I cannot expect the system I work > with to conform to the standard? I have not seen any Unix-like system in decades which did not have standard man pages available. Some may not install them by default, just as some might not install a compiler by default. But I accept the risk that a system will be incomplete. You will note that the vast majority of people who install OS X never install a compiler for it. Not every system that has a compiler has gcc. And yet, I consistently find that if I write code built for C89, it can be gotten to work nearly anywhere, but if I write for GNU C, I run into more hiccups. So I write for standard C as much as I can. Which is sometimes not very much, although I'm a lot more likely to depend on POSIX, or even on Linux specifically, than on GNU language extensions. Anyway, it's up to you. You can write whatever kind of documentation you want, for whatever tools you want. You can use texinfo, you can use man, you can use POD, you can use RDoc, whatever pleases you. My observation is merely that, if I find a command in my path, the thing I'm going to try first to figure out how to use it is "man <command>", and if that doesn't work, I tend to be distrustful of the code. So far, that's worked pretty well for me as a heuristic, even though I can, of course, identify plenty of counterexamples. -s -- Copyright 2010, all wrongs reversed. Peter Seebach / usenet-nospam(a)seebs.net http://www.seebs.net/log/ <-- lawsuits, religion, and funny pictures http://en.wikipedia.org/wiki/Fair_Game_(Scientology) <-- get educated!
From: Ivan Shmakov on 4 Feb 2010 16:51 >>>>> "S" == Seebs <usenet-nospam(a)seebs.net> writes: [...] S> My observation is merely that, if I find a command in my path, the S> thing I'm going to try first to figure out how to use it is "man S> <command>", and if that doesn't work, I tend to be distrustful of S> the code. So far, that's worked pretty well for me as a heuristic, S> even though I can, of course, identify plenty of counterexamples. Agreed. Though personally I tend to use $ COMMAND --help first. -- FSF associate member #7257
From: Janis Papanagnou on 4 Feb 2010 17:11 Ivan Shmakov wrote: > > Agreed. Though personally I tend to use $ COMMAND --help first. > How well does COMMAND --help work on current Unixes (AIX, HP-UX, Solaris, etc.)? When I used those systems (and others like UTS, old SunOS, etc.) in the past they don't seem to have supported --OPTION at all, AFAIR. If some admin, at that time, abstained from installing the man pages to save some disk space, I got very upset, I recall. The man pages were always extremely helpful, and the fastest access to the most relevant information on all Unix systems I used. Janis
From: Stephane CHAZELAS on 4 Feb 2010 17:22
2010-02-04, 23:11(+01), Janis Papanagnou: > Ivan Shmakov wrote: >> >> Agreed. Though personally I tend to use $ COMMAND --help first. >> > > How well does COMMAND --help work on current Unixes (AIX, HP-UX, Solaris, > etc.)? When I used those systems (and others like UTS, old SunOS, etc.) in > the past they don't seem to have supported --OPTION at all, AFAIR. [...] a useful one is "COMMAND -:" (":" can't be a valid option for commands using getopt, so you're likely to get an error message and a usage summary). Alternatively, you can use -\? but it's longer to type. whatis or man -f, strings... -- St�phane |