I have this variable called FirstIN that contains something like this: 001,002,003,004...
I am trying to assign the content of this variable into ModifiedIN but with the following format : 001 002 003 004...(changing the commas for spaces)
I thought about using sed but i am not getting the expected result...which i guess is due to my misunderstanding of the command...
Maybe with tr instead? How would i copy the content of the original var into the second var while using tr? tried ModifiedIN=$FirstIN | tr ',' ' ' but that is wrong...
It's not a bash-ism. It works fine in a bourne shell. I even tested it in busybox ash, one of the most limited, stripped-down, barebones bourne shells still available today.
That would explain why it even works in vanilla ksh when I finally caved in and installed it. Solaris still ships with non-POSIX shells? How lovely to get accused of BASH-ism for following POSIX...
Where can I find such a truly garbage shell to test with, to ensure code we suggest will work with solaris? :rolleyes:
Even the sun guys find it odd but in Solaris backward compatibility is king, so taking anything out once it's there is hard. There are POSIX shells on Solaris also but probably the most common environment that I see for a POSIX shell snippet not working in a given environment is Solaris.
KSH93 supports it (perhaps it originated there?), but KSH88 does not. I suspect that SFNYC, like myself, is using pdksh which is KSH88 for the most part.
For the sticklers, it still hasn't found its way into the POSIX sh standard.
Older ksh88 version is not Posix compatible, because posix has done later. Most of commercial *nix still include only ksh88 version.
But you can download also ksh93 version, which is posix-compatible. Why ? I think that Posix-team has used lot of ksh93 to make posix standard. Download binary.
Simple test, ksh88 or ksh93. Many *sh based on ksh88 ...
Ex. substr not work in older shells.
I don't understand how you come to the conclusion that ksh93 is somehow more posix-y than ksh88. If you take a look at the features it introduced (associative arrays, floating point arithmetic, new forms of parameter expansion, etc), 15 years later they're not part of the posix sh standard. If anything, ksh93 deviates further from the standard.
The example you chose, substring parameter expansion, exemplifies this. It is a feature that's supported by most modern variants of the bourne shell, but it is not part of posix. Any shell that does not implement it is not any less posix-compliant because of it. The same goes for the pattern/replacement parameter expansion that started this discussion. (Though they are both handy features, no doubt.)
You are right. Shell Command Language
I didn't check... I have learned this substr feature, not from ksh, it was some "posix-sh". Now I know, it wasn't :). Or it was - include posix.
Dash is maybe the most of posix-compatible and only posix. I know that ksh and many other "Bourne shell child" include lot of extension features. Much more than posix shell. For me posix shell compatible means that sh include posix. I try to use only posix stuff (almost :).
Anyway, parameter extension is powerful, but not so easy to read. Need always little comments for others, who are not so sh gurus. sed, tr, cut, awk, expr, .. lines are more readable, but need huge cpu if compare using of parameter extension. Not built-in.
So, my test was simple test to compare, have you ksh88 or ksh93. Nothing else.
