Re: GNU Awk - inplace editing

On 23.05.2025 10:58, Kaz Kylheku wrote:

On 2025-05-23, Kenny McCormack <gazelle@shell.xmission.com> wrote:

In article <100p11m$3uh3m$1@dont-email.me>,
Janis Papanagnou  <janis_papanagnou+ng@hotmail.com> wrote:

In GNU Awk I was looking for the in-place option (similar to sed -i).
I thought there once was some _simple_ option usable from the command
line. (Or am I misremembering?)
>
The manual now suggests to use a GNU Awk "inplace" _Extension_ for that
 gawk -i inplace ...
and
 gawk -i inplace -v inplace::suffix=.bak ...
respectively.
>
That's not exactly as simple to use as, say,
 gawk -i ...
and
 gawk -i.bak
so I suppose there's a reason for the added complexity in the handling.
>
Does anyone know that reason or remember a rationale? - I don't recall
any discussions about that...

>
I've explained this a few times over the years (in this newsgroup).

It's no disrespect; my memory is just limited.

>
There was never a "-i" option in Gawk that meant "inplace" (and there never
will be).
>
The key to understanding this is to understand that (in Gawk), the "i" in
"-i" does not stand for "inplace".  It stands for "include".

No doubt. - What I wanted to say is that I thought there was a _simple_
option "like -i" (although, before 'include' times, it could have also
been '-i').

>
Once you understand that, all becomes clear.

Unfortunately, the question about the rationale for using the GNU Awk
Extension mechanism - which was my original question - isn't answered
by that or any clearer.

 
Sure, but, interestingly, just like Janis, I also seem to have a false,
memory of there having been some other inplace mechanism that was
replaced by the -i inplace include (not necessarily a -i option).

I never needed the "inplace" feature with Awk[*] so my (wrong) memories
were not based on practical experience; practical experience would have
lead to a more enduring (likely correct) memory, I'd expect.

[*] On shell level I use a two-step process with a temporary, to make
processing typically also a bit more reliable.

There is no evidence of any such in the available materials, though.
 
We might have both been duped by something unclear someone said once,
(perhaps here)?

It's no biggie, I just wanted some clarity.

Thanks for the investigation.

Janis