Re: gensub() matched component always empty, even for the exact text of the example in the manual

[Ed Morton]

It's unlikely that a bug like that could exist for all this time without 
hundreds/thousands of people reporting it. I don't have that gawk 
version any more but I don't see it in gawk 5.1.1 on cygwin:

   $ gawk --version | head -1
   GNU Awk 5.1.1, API: 3.1 (GNU MPFR 4.1.0, GNU MP 6.2.1)

   $ gawk '
   BEGIN {
         a = "abc def"
         b = gensub(/(.+) (.+)/, "\\2 \\1", "g", a)
         print b
   }'
   def abc

I suspect you copy/pasted the example wrong. Maybe you re-typed it 
instead of copying it and your editor used some other chars instead of 
the ASCII ones required?

Regards,

    Ed.


On 4/3/2022 8:29 AM, Zoli Márk via Bug reports only for gawk. wrote:
> Hello -
> I have run into a problem for which I needed the (g)awk function gensub(), namely the 
> "identify matched components and reuse them in the replacement text" kind of 
> problem. However, it seems that the component thus matched is always empty, and 
> investigating the issue has shown that in this particular gawk not even the manual 
> example works.
> This happens on Ubuntu 20.04.
> gawk -V replies "GNU Awk 5.0.1, API: 2.0 (GNU MPFR 4.0.2, GNU MP 6.2.0)" plus 
> boilerplate. I can't tell anymore if I installed it from some repository or it came with 
> Ubuntu - I am, however, sure that I did not install it from sources.
> The manual I'm referring to is the one linked from thehttps://www.gnu.org/software/gawk/  page, namelyhttps://www.gnu.org/software/gawk/manual/gawk.html  
> The example is the following:
> $ gawk '
> > BEGIN {
> >       a = "abc def"
> >       b = gensub(/(.+) (.+)/, "\\2 \\1", "g", a)
> >       print b
> > }'
> -| def abc
> Instead of printing the promised "def abc", on my computer this code prints an 
> empty line only, as both \\1 and \\2 are empty. (Actually it's not an empty line - it 
> contains the space from between \\2 and \\1 in the replacement text, as it did with other 
> characters in my code; only the matched components themselves are empty.) Is this a bug?
> Best regards
>         M- S- Z-