Shell Programming and Scripting

$((m++))

That's not portable (i.e., POSIX); it is not a valid Unix shell statement.
It may work in some shells, but there's no guarantee.

The reliable way is:

Code:

$(( m += 1 ))

No. There is no good reason to write unportable syntax when the portable is just as efficient.

True, and they should only be used when the portable method is less efficient.

The vast majority of scripts can be written using only POSIX syntax.

that's my illustration of being not standard, but yet is also a valid syntax provided by bash.

Quote:

The reliable way is:

Code:

$(( m += 1 ))

don't work in bourne shell

Quote:

No. There is no good reason to write unportable syntax when the portable is just as efficient.

not talking about portable or not portable here. Bash provides the facility to write non portable code such as the syntax I described, but are you going to stop every user who bash not to write it that way?

Quote:

True, and they should only be used when the portable method is less efficient.

efficient? who cares (except cfajohnson)? are you going to next show me proof that let "m+=1" is slower or faster the $((m+1)) ?

Quote:

The vast majority of scripts can be written using only POSIX syntax.

If there are only one type of shell in this world, who needs a POSIX standard? while I do not disagree with you on this, not all scripters want to write in POSIX mode as well. A vast majority of them prefers to utilize their shell to the fullest to get the job done.

IOW, it is limited to bash. Why do that when you can write in a way that works in bash as well as all other POSIX shells?

Quote:

don't work in bourne shell

Nor does any other form of arithmetic. But the Bourne shell is not a standard Unix shell (it is a precursor to it).

It is much faster when you have to rewrite your script for a standard shell that doesn't support that extension.

Quote:

If there are only one type of shell in this world, who needs a POSIX standard?

There are many shells; that's why the standard is important.

Quote:

while I do not disagree with you on this, not all scripters want to write in POSIX mode as well. A vast majority of them prefers to utilize their shell to the fullest to get the job done.

I, too, utilize the shell to the fullest. The way I do that is to use the most portable syntax available. When I need something that is not available in the standard shell, then and only then do I use the more parochial syntax.

Quote:

Nor does any other form of arithmetic. But the Bourne shell is not a standard Unix shell (it is a precursor to it).

You said the more reliable way is $((m+=1)). I said it does not work in bourne shell. But if you want to talk about POSIX and stuff and if I am still using the bourne shell or tcsh, then it will not work right? In any case, this is also OT

I know why the standard comes about. You did not get my point and I am too lazy to explain my point to you either.

Quote:

I, too, utilize the shell to the fullest. The way I do that is to use the most portable syntax available.
When I need something that is not available in the standard shell, then and only then do I use the more parochial syntax.

so just to get this clear, you do write in a way that is usable in bourne shell, zsh , tcsh , dash, etc, correct?

Isn't that what I said? When I want to do something that is more efficiently done using a non-portable syntax, I do that. (BTW, you can thank me for suggesting the ${var^} and ${var,} expansions to Chet Ramey.)

However, if I am unable to use bash4 (it is not installed on some systems I use) I don't use external commands for those conversions; I use functions that use only the shell.

I write, as much as possible, in a way that will run in any standard Unix shell (that is, a POSIX shell).