1:09 p.m.
Is it possible to set buffer-file-coding-system-for-read to a coding
system such that:
1. Loading a file, modifying it, and saving it does not change its EOL
type.
2. Line breaks are auto-detected (and displayed without control
charatcers etc).
?
I find that 'no-coding satisfies 1 but I have to live with ^M's, and
that every other coding system I try uses a particular, fixed, eol
type for write. I expected 'undecided to be the answer, but this
strips CR's from CRLF encoded files on write.
FWIW:
XEmacs 21.5 (beta9) "brussels sprouts" [Lucid] (i686-pc-cygwin) of
Thu Sep 26 2002 on PASTON
CYGWIN_NT-5.0 PASTON 1.3.12(0.54/3/2) 2002-07-06 02:16 i686 unknown
Thanks,
- Phil
1:17 p.m.
Philip Aston wrote:
Is it possible to set buffer-file-coding-system-for-read to a coding
system such that:
1. Loading a file, modifying it, and saving it does not change its EOL
type.
2. Line breaks are auto-detected (and displayed without control
charatcers etc).
?
I find that 'no-coding satisfies 1 but I have to live with ^M's, and
that every other coding system I try uses a particular, fixed, eol
type for write. I expected 'undecided to be the answer, but this
strips CR's from CRLF encoded files on write.
As the name implies, buffer-file-coding-system-for-read only applies
to reading the file. buffer-file-coding-system determines the EOL
convention used when writing the file.
--
Glynn Clements <glynn.clements(a)virgin.net>
1:32 p.m.
Glynn Clements writes:
Philip Aston wrote:
> Is it possible to set buffer-file-coding-system-for-read to a coding
> system such that:
>
> 1. Loading a file, modifying it, and saving it does not change its EOL
> type.
>
> 2. Line breaks are auto-detected (and displayed without control
> charatcers etc).
>
> ?
>
>
> I find that 'no-coding satisfies 1 but I have to live with ^M's, and
> that every other coding system I try uses a particular, fixed, eol
> type for write. I expected 'undecided to be the answer, but this
> strips CR's from CRLF encoded files on write.
As the name implies, buffer-file-coding-system-for-read only applies
to reading the file. buffer-file-coding-system determines the EOL
convention used when writing the file.
Thanks.
So.. urmm.. same question but
s/buffer-file-coding-system-for-read/buffer-file-coding-system/g ?
Namely, what do I set (or rather set-default) it to such that eol type
is preserved?
- Phil
2:22 p.m.
Philip Aston wrote:
> > Is it possible to set buffer-file-coding-system-for-read
to a coding
> > system such that:
> >
> > 1. Loading a file, modifying it, and saving it does not change its EOL
> > type.
> >
> > 2. Line breaks are auto-detected (and displayed without control
> > charatcers etc).
> >
> > ?
> >
> >
> > I find that 'no-coding satisfies 1 but I have to live with ^M's, and
> > that every other coding system I try uses a particular, fixed, eol
> > type for write. I expected 'undecided to be the answer, but this
> > strips CR's from CRLF encoded files on write.
>
> As the name implies, buffer-file-coding-system-for-read only applies
> to reading the file. buffer-file-coding-system determines the EOL
> convention used when writing the file.
So.. urmm.. same question but
s/buffer-file-coding-system-for-read/buffer-file-coding-system/g ?
Namely, what do I set (or rather set-default) it to such that eol type
is preserved?
Almost anything that doesn't have a suffix of -unix, -dos or -mac
(except "binary"). The exact choice depends upon how you want
non-ASCII characters to be handled by a MULE-enabled version of
XEmacs.
Basically, every coding system has four variants. E.g. for ISO-8859-1,
you have: iso-8859-1, iso-8859-1-unix, iso-8859-1-dos and
iso-8859-1-mac. The "basic" version autodetects the EOL convention,
while the others assume a specific convention (LF, CRLF, and CR
respectively).
If you wish to autodect the base coding system, as well as the EOL
convention, use "undecided".
--
Glynn Clements <glynn.clements(a)virgin.net>
7:43 a.m.
Glynn Clements writes:
Philip Aston wrote:
> So.. urmm.. same question but
> s/buffer-file-coding-system-for-read/buffer-file-coding-system/g?
>
> Namely, what do I set (or rather set-default) it to such that eol
> type is preserved?
Almost anything that doesn't have a suffix of -unix, -dos or -mac
(except "binary"). The exact choice depends upon how you want
non-ASCII characters to be handled by a MULE-enabled version of
XEmacs.
With both buffer-file-coding-system and
buffer-file-coding-system-for-read set to "raw-text" I can read in a
CRLF encoded text file (which displays OK), but writing it changes the
file to CR format. M-x toggle-buffer-file-coding-system and then
writing saves in CRLF format (which is why I suspected
buffer-file-coding-system-for-read in the first place).
Basically, every coding system has four variants. E.g. for
ISO-8859-1, you have: iso-8859-1, iso-8859-1-unix, iso-8859-1-dos
and iso-8859-1-mac. The "basic" version autodetects the EOL
convention, while the others assume a specific convention (LF,
CRLF, and CR respectively).
If you wish to autodect the base coding system, as well as the EOL
convention, use "undecided".
Same behaviour as "raw-text", but with the addition of exiting xemacs
process when I start ispell :-(
Persisting I tried the following:
(setq buffer-file-coding-system-for-read 'undecided)
(setq-default buffer-file-coding-system 'raw-text)
Then with a CRLF text file (test.txt) I see the following:
C-x f test.txt
C-h v buffer-file-coding-system =
Value: #<coding-system raw-text-dos no-conversion eol-type=crlf>
Default-value: raw-text
(Good)
Change the file, then C-x s test.txt
C-x k
C-x f test.txt
C-h v buffer-file-coding-system =
Value: #<coding-system raw-text-dos no-conversion eol-type=crlf>
Default-value: raw-text
(Good)
C-x k
C-x d
C-x f test.txt
C-h v buffer-file-coding-system =
Value: #<coding-system raw-text-unix no-conversion eol-type=lf>
Default-value: raw-text
HUH?
How/why does dired screw up the buffer-file-coding-system used for
reading test.txt?
AFAICT it doesn't modify buffer-file-coding-system-for-read, the
default value of buffer-file-coding-system, file-coding-system-alist,
find-file-hooks, insert-file-contents-pre-hook,
insert-file-contents-post-hook, ... What am I missing?
- Phil
4:06 p.m.
How/why does dired screw up the buffer-file-coding-system used for
reading test.txt?
AFAICT it doesn't modify buffer-file-coding-system-for-read, the
default value of buffer-file-coding-system, file-coding-system-alist,
find-file-hooks, insert-file-contents-pre-hook,
insert-file-contents-post-hook, ... What am I missing?
I think there are some lurking bugs. I tried doing some stuff with custom
coding systems and it didn't work at all.
andy
8087
days inactive
8088
days old
5 comments
3 participants
participants (3)
-
Andy Piper -
Glynn Clements -
Philip Aston