i have a windows powershell script i need to adapt
to run on a linux system. in it there are a number
of double precision variables. is that something
that can be done within a bash script or would i
need to move to somethig like c/c++?

if i can use double precision variables in bash,
how do i declare the variable? haven't seen that.

thanks,
BabaG

I don't know that bash has a notion of double precision variables,
but you can assign arbitrary values to variables, and use 'bc' to
manipulate those.


Cheers,
Tink

If the script is not too complex you could translate it into a PHP or Perl script where it will be much easier to operate on double precision values. Both will work from the commandline (PHP needs to be cli-enabled).

KDE Programs Naming Convention

thanks for the replies. bc sounds like the way to
go for my purposes. a question, though.

i have a text file holding a lot of values. it has
been read into an array. some of the elements are
strings and some are numerical. of the numerical
ones, some are integer, some fractional. i'm having
trouble finding how i'd declare a fractional variable
from this array for piping to bc.

$GamInc=.5

if i say:

echo "$GamInc*5" | bc

i get an error, rather than the 2.5 i'd expect.
how do i get around this?

thanks again,
BabaG

When setting variables, don't use a "$":
The ~$ is my prompt.

actually, to give more detail, i have my script read the
text file into the array. each line of text becomes an
element of the array. i then am assigning individual
variable names to each element so i can follow them more
clearly.

in the first gathering from the elements from the text
file, the array is called ScriptVariable[n].

i then perform a series of re-assignments, such as:

GamInc="{$ScriptVariable[75]}"

if i then say:

echo "$GamInc*5" | bc

it produces the following error:

(standard_in) 1: illegal character: ^M

something seems to be happening when i do the re-assigning
of names. if i simply echo the variable to the screen
(echo $GamInc), it looks ok. if i try to perform a
mathematical operation, it errors.

thanks,
BabaG

You wrote your script in windows, didn't you :}


Try
and then work with newscript instead ...

Cheers,
Tink

thanks tinkster. could you elaborate on why i'd use the fromdos
command? never heard of it till now.

as i said at the start, this started as a powershell script.
i'm assuming there will be a lot of changes to convert to a
bash script. yes, the powershell version was written in windows.
i'm not directly copying it to bash, however. i have a number of
practice bash scripts for various things like reading text file
input to an array. i've copied parts of that to make this
re-written bash version. i've also combined that initial reading
of the text file with the renaming section from the powershell
script by pasting and changing some characters from the
powershell version to conform to bash syntax.

so, while parts of this did, in fact, start in windows, the
actual script being discussed at present is something of a
hybrid, mostly originating in bash. and, as i say, the echoing
to the screen works fine, it's just the piping to bc that's
erroring.

would fromdos still be appropriate in an instance like i've
described here?

thanks,
BabaG

fromdos/todos (or dos2unix/unix2dos in some distros)
are quite simple tools that replace the line-ends for
use in the other OS. Unix uses a plain LF, DOS (Windows)
uses a CR/LF ... the CR is what you see as ^M in your
scripts warning message. If they're not installed you
could use a simple sed-script to do that translation...

I'd think so, since you're getting the error pertaining
to DOS line-ends.


Cheers,
Tink

thanks. looking for those now. first try produced
'command not found' for both fromdos and dos2unix.
will look for an rpm. (mandriva 2007)

thanks again,
BabaG

As I said ... sed can do that, too.

Probably quicker than a download... ;}



Cheers,
Tink