Re: sed: Case-insensitive matching does not work

[Francesco Potorti`]

> /foo/Id        deletes lines containning the regex, case insensitive;
> s/foo/bar/I    replacement---``foo'' is interpreted case insensitive;
> 
> /xxx/Is/foo/bar/   replacement which is performed on each line matching
>                    the /xxx/ interpreted case insesitively; the /foo/
>                    is interpreted case sensitively.
> 
> /xxx/s/foo/bar/I   replacement is performed on lines matching /xxx/ (case
>                    sensitive), the replacement itself is case insensitive.

Hmmm.  Okay.  I had missed this point.  Thank you.

> These two bits of information are independent, even when you use the
> shortcut
>       /foo/s//bar/

I think this is correct and consistent.  Only it should be stressed in
the manual, while as far as I know it is not even mentioned.

>       s/href=raw/href=pot/I
> 
> I consider the last form as the most readable one.
> Is there any special reason why it cannot be used?

No, in fact it can be written this way.  I tend to begin all my sed
instructions with a regexp qualifier, and I find that more readable, but
then I have just learnt about the case-insensitive matching capabilities
of GNU sed, so I may need to review my habits.

Thanks again.

I think the bug should be downgraded to a wishlist asking for
documentation about this issue.