Re: What's wrong with this code??

Re: What's wrong with this code??

[Santanu]

On May 18, 2:20 am, Andreas Politz <poli...@fh-trier.de> wrote:
> Santanu <thisissant...@gmail.com> writes:
> > Hello Everybody,
>
> > Can you tell me what is wrong with this code:
> > -------------------
> > (setq
> >    go-packages-installed
> >    (split-string
> >     (shell-command-to-string
> >      "cd $GOROOT/src/pkg && find . -iname *.go | sed -e 's|^\./||g; s|/
> > [^/]*\.go||g' | sort | uniq"\
> > )))
>
> You have to escape the backslash characters, because of string
> expansion.

Spot on. Thank you very much.
BTW, can you also explain a bit what it means when I don't put the
double backslash character, and instead use a single backslash?
I ask this because my original code (the incorrect one above) did
produce some correct results.

Regards,
Santanu

Re: What's wrong with this code??

[Frédéric Perrin]

Santanu <thisissantanu@gmail.com> writes:
> On May 18, 2:20 am, Andreas Politz <poli...@fh-trier.de> wrote:
>> Santanu <thisissant...@gmail.com> writes:
>> > Can you tell me what is wrong with this code:
>> > -------------------
>> > (setq
>> >    go-packages-installed
>> >    (split-string
>> >     (shell-command-to-string
>> >      "cd $GOROOT/src/pkg && find . -iname *.go | sed -e 's|^\./||g; s|/
>> > [^/]*\.go||g' | sort | uniq"\
>> > )))
>>
>> You have to escape the backslash characters, because of string
>> expansion.
>
> BTW, can you also explain a bit what it means when I don't put the
> double backslash character, and instead use a single backslash? I
> ask this because my original code (the incorrect one above) did
> produce some correct results.

The first expression in the sed was « s|^./||g; » (which didn't hurt,
because it seems that you don't have directories whoose name contains
only a single letter), and the second wes « s|/[^/]*.go||g », with the
LHS matching strings such as « /svgo ».

-- 
Fred

Re: What's wrong with this code??

[Santanu]

On May 18, 7:37 pm, Frédéric Perrin  <fred...@SPAM.resel.fr> wrote:
> Santanu <thisissant...@gmail.com> writes:
> > On May 18, 2:20 am, Andreas Politz <poli...@fh-trier.de> wrote:
> >> Santanu <thisissant...@gmail.com> writes:
> >> > Can you tell me what is wrong with this code:
> >> > -------------------
> >> > (setq
> >> >    go-packages-installed
> >> >    (split-string
> >> >     (shell-command-to-string
> >> >      "cd $GOROOT/src/pkg && find . -iname *.go | sed -e 's|^\./||g; s|/
> >> > [^/]*\.go||g' | sort | uniq"\
> >> > )))
>
> >> You have to escape the backslash characters, because of string
> >> expansion.
>
> > BTW, can you also explain a bit what it means when I don't put the
> > double backslash character, and instead use a single backslash? I
> > ask this because my original code (the incorrect one above) did
> > produce some correct results.
>
> The first expression in the sed was « s|^./||g; » (which didn't hurt,
> because it seems that you don't have directories whoose name contains
> only a single letter), and the second wes « s|/[^/]*.go||g », with the
> LHS matching strings such as « /svgo ».

Ah... I see. Now I understand. So, basically, just a singe '\'
character in an elisp
regexp string acts as if it wasn't even there in the first place.

Thank you for the explanation.

Regards,
Santanu

Re: What's wrong with this code??

[Stefan Monnier]

> Ah... I see. Now I understand. So, basically, just a single '\'
> character in an elisp regexp string acts as if it wasn't even there in
> the first place.

Yes, except when it's a valid escape sequence, i.e. when the backslash
is followed by r, t, ^, C, M, 0-9, x, o, d, e, a, b, v, LF, n, ", s, u,
U, or f (and maybe a few more ;-).


        Stefan

Re: What's wrong with this code??

[Santanu]

On May 19, 2:04 am, Stefan Monnier <monn...@iro.umontreal.ca> wrote:
> > Ah... I see. Now I understand. So, basically, just a single '\'
> > character in an elisp regexp string acts as if it wasn't even there in
> > the first place.
>
> Yes, except when it's a valid escape sequence, i.e. when the backslash
> is followed by r, t, ^, C, M, 0-9, x, o, d, e, a, b, v, LF, n, ", s, u,
> U, or f (and maybe a few more ;-).

Thanks. I will have to keep this in mind too.
Things now got a bit more complex :-)

Regards,
Santanu