Kai == Kai Großjohann [EMAIL PROTECTED] writes:
On Fri, 25 May 2001, Sven Utcke wrote:
(Should Tramp assume that any Perl found is a Perl 5?)
Hmm. This, of course, would be guaranteed to break once perl6 is
out, and possibly on some (very few) perl4-only machines...
I guess it's time to
Sven == Sven Utcke [EMAIL PROTECTED] writes:
But the strange thing of course is that it _does_ work with that
particular instance of perl when called from the shell, just not when
called from tramp. This really _is_ weird...
Could it be something odd going on with the line-endings ?
"Pete" == Pete Forman [EMAIL PROTECTED] writes:
Stefan Monnier writes:
Just guessing. I refuse to believe that someone would write
a shell whose history file feature cannot be turned off.
The POSIX/UNIX rules for shells mandate that history cannot be turned
off, unless the shell
"Pete" == Pete Forman [EMAIL PROTECTED] writes:
It is moot as to whether /dev/null is suitable for HISTFILE. While it
does have read and write permissions I would infer that the shell
expects to be able to read back what is written. A robust
implementation should be able to cope but can we
"Ted" == Ted Stern [EMAIL PROTECTED] writes:
/usr/bin/env is available on Solaris, at least.
Experience shows you can't rely on it being available.
And in any case it's not much better than
`ksh -c "HISTFILE=/dev/null exec ksh"'
Stefan
For example, what happens if we do "HISTSIZE=0" in addition to
"unset HISTFILE"?
IIRC I tried that, and it didn't work: ksh made temp files.
Have you tried /dev/null ?
Also since the current code doesn't use here-documents any more, your
.sh_history file should grow much slower now (it
[EMAIL PROTECTED] writes:
In my experience setting HISTFILE=/dev/null did not work: ksh merely
made a tempfile.
Of course, this still uses up disk space, but now that only actual commands
rather than file contents go into the history file, it's less of a concern.
The remaining question is:
I think that turning off ksh's history is a minor issue which can be
handled is various ways
Really? Actually, there was a thread on that topic on this list
recently. If you have a solution, I'd appreciate hearing it. Or are
you just guessing?
Just guessing. I refuse to believe that
"Daniel" == Daniel Pittman [EMAIL PROTECTED] writes:
Basically, the admin command was passed
-i/r@METHOD:HOST:/PATH/TO/FILE/ where -iFILE would suffice.
Compounding the problem, the error wasn't noticed and the file and
buffer destroyed.
I only have the development version of VC right now,
[EMAIL PROTECTED] writes:
@@ -177,7 +177,7 @@
(apply 'vc-do-command nil 0 "admin" nil
(and rev (concat "-r" rev))
"-fb"
- (concat "-i" file)
+ (concat "-i" (file-relative-name file))
(and comment (concat "-y" comment))
-
"Benjamin" == Benjamin Rutt [EMAIL PROTECTED] writes:
I have not frobbed the backup-by-copying variable or any others. The
default value appears to be nil on my system, which would suggest that
renaming is used (and therefore modes should be preserved).
Indeed, renaming is then used to
works for me, and seems even cleaner. (Except for the paren placement:
being usually procedural I like starts and ends to line up :-)
There are coding conventions in most/all languages. It's usually considered
good practice to follow them, even if your habits disagree at first.
"Daniel" == Daniel Pittman [EMAIL PROTECTED] writes:
Point of view, really. rest accepts zero or more arguments, optional
accepts zero or one. It makes little difference if you only ignore them.
But since the argument is ignored and since no argument is passed when
used interactively (since
Yes. Despite `file-remote-p'[1] being available, many packages are
hard-coded to the knowledge of FTP-style paths.
VC did (in Emacs-20) and ediff still does.
Which others ?
Yes. Especially because, for example, VC will refuse to operate on a FTP
file path (because it can't), while TRAMP
"Daniel" == Daniel Pittman [EMAIL PROTECTED] writes:
Isn't C-u [1] C-x C-f what you want there, or is that just XEmacs?
I think it's an Emacsism (but I wouldn't know).
It prompts me interactively for a coding system to visit with...
Is that the coding system used for the file's content ?
I
"Francesco" == Francesco Potorti` [EMAIL PROTECTED] writes:
Emacs should have a general hook for /[^/]+:.* filenames, where [^/]+ is
the protocol. Then, different packages could register to that hook and
tell it which protocol they do manage.
I think using /protocol: is just fine, indeed.
"Pete" == Pete Forman [EMAIL PROTECTED] writes:
/![@enc]/:telnet://[usr[:pwd]@]host1[:port]/ssh://host2/:/path/to/file
[...]
F tramp-prefix-authority"//:" or "#"
Clearly ""//:"" won't do since Emacs tends to interpret it directly.
I liked the @enc thing, except that I'd use it for the
"Kai" == Kai Grojohann [EMAIL PROTECTED] writes:
Normally, you can do things like this:
M-! ghostview ~/foo.ps RET
Note the ampersand. It does what you think it does: the same as on a shell.
Have you tried M-! (gv ~/foo.ps ) RET ?
It should DTRT, but it might also seriously screw up
"Kai" == Kai Grojohann [EMAIL PROTECTED] writes:
I have suggested a remedy once before, and the end result was that RMS
suggested to create a new filename handler called `process-file'.
Then Tramp can implement this file operation, too, and then
`compile-internal' can be rewritten to use the
"Stefan Monnier" 06 Mar 2001 15:15:53 -0500
Last we talked about it, my argument about `start-process' not being
fundamentally different from `shell-command' was rather well
received, which leads me to believe that it might be possible to
convince the maintainers that mak
"Henrik" == Henrik Holm [EMAIL PROTECTED] writes:
since I know very little about cvs, I've let Emacs/pcl-cvs
handle updating from the CVS repository. does pcl-cvs have
any command for updating to a specific version, or a specific
branch (and is the 2.0.x.x branch the one I should pursue?)
"Daniel" == Daniel Pittman [EMAIL PROTECTED] writes:
like? On a branch, of course, because a non-working version as the trunk
would be /very/ bad. :)
Why would it be bad ?
Just make sure there is a `stable' tag pointing to the last working
release and you're set.
Stefan
Branches are not pure evil, but I'd recommend against using them in
this case, because they will probably bring more hassle then they
deserve.
I think branches are a problem for small development teams. Since the
Tramp development team is very small (a quarter of a Kai plus Daniel),
"Kai" == Kai Grojohann [EMAIL PROTECTED] writes:
I'm thinking that this might be a good time to change the revision number to
2, and to create a STABLE branch on CVS. Then people can start to implement
new features in the head branch without impacting the users.
Branches are not pure evil,
[EMAIL PROTECTED] writes:
I noticed that you hardwired TERM to be "dumb". If you are interested,
here is a simple patch that allows the user to set TERM via
tramp-terminal-type. I personally use "tramp" because then I can catch
that in my .zlogin and make sure I don't prompt for a terminal
"Skip" == Skip Montanaro [EMAIL PROTECTED] writes:
I'm still generally unable to use ediff with tramp file specs (I asked back
Does `ediff-buffers' work ?
Stefan
"James" == James Cooley [EMAIL PROTECTED] writes:
while I cannot fetch the files because rcp does not read my .profile. I'm not
sure if there is an answer to this,
Isn't there some way to make this `rcp' access difficult to use
and make the inline method "in your face" so that those problems
"Harry" == Harry Putnam [EMAIL PROTECTED] writes:
With my lisp skill level that will probably happen at about the same
time they start selling `snowcones' in the hot place. But I
So sad! Elisp is very easy and with Edebug it's a real pleasure to
write and extend code. I'd warmly encourage
This is a very good point. So, you're saying that ssh should _not_ be
prompting for both passphrase and password, correct? I need to confirm
Not really. Just that SSH, not being able to know in advance whether your
private identity-key can be used to login, first tries to log in
with it and
"Skip" == Skip Montanaro [EMAIL PROTECTED] writes:
If you want to include a "]" or a "-" inside a set, precede it with a
backslash, or place it as the first character.
Note that Emacs' syntax for [...] treats backslashes normally, so they
can't be used for quoting. This again follows
"Jeffrey" == Jeffrey Juliano [EMAIL PROTECTED] writes:
I'm far from an expert here, but seems to me that one would have to
modify remote-compile in order to get it to work for you. The
modifications necessary to get it to hop through multiple firewalls
might look very similar to code in
What is `c'?
An archaic expression meaning 'and such' or, more verbosely, 'and more
things like that'. Comes from printing habits in England around 1900 or so.
If you take as "and" and translate it into some kind of latin-derived
language like French, you get "et". Concatenation with the
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
"find . \! -name . -prune -print" does not recurse, but "find . \!
-name . -prune -name b\* -print" does recurse. Why?
Stefan, Oh Great Find Guru?
I'm no find guru, but it indeed seems to be hitting yet-another-find-bug.
[ GNU find has been
"Owns" == Owns all emacs-rcp files in CVS [EMAIL PROTECTED]
writes:
Make `tramp-line-end-position' more useful - it now knows the XEmacs name for
'what is the point at the end of the line,' and actually uses it as well.
Rather than
(defun f () (cond ((fbound bla) bli) ((fbound foo)
"Joe" == Joe Stoy [EMAIL PROTECTED] writes:
2) avoiding automount failures for irrelevant directories.
That's a very good point.
It seems that `find . \! -name . -prune -print' is clever enough not to
stat every file it finds, but `find . -type d \\! -name . -prune -print'
isn't that clever.
I
a) ls -a 2/dev/null | f?grep ^%s
Of course, it's better to use `grep ^%s' than `fgrep %s' since it solves
both the leading dash and the substring issue, but then you have to
carefully requote your string.
Maybe something like:
(format "%s -a 2/dev/null | grep ^%s"
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
I'm not sure if it is possible to fix this easily. Hm. Maybe I could
`ls -l' the whole directory and then pick out the information that's
needed from that. Yes, that could work. But it might be
inefficient. I'll put it on the todo list
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
Well, the precedence (with w3 and ffap) seems to be to provide
commands tramp-bug and tramp-submit-bug. I'll do that.
^
You forgot-report
At least, based on M-x apropos RET
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
"Stefan Monnier" [EMAIL PROTECTED] writes:
There are many alternatives. For example using `ls -aF' instead of `ls -a'
would immediately give the directory info (ideal).
The `-F' argument for `ls' is too no
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
But it seems to me that the _right_ name would be report-tramp-bug,
wouldn't it? (And report-gnus-bug and report-cc-mode-bug and so on?)
I don't know about the "right" name. It's still considered bad practice
to use global identifiers that
"Francesco" == Francesco Potorti` [EMAIL PROTECTED] writes:
M-x report SPC SPC and you are given a list from which you choose.
You mean M-x repor TAB RET ?
My Emacs doesn't give me any such list, is that an XEmacs thing ?
Stefan
"Daniel" == Daniel Pittman [EMAIL PROTECTED] writes:
* Get group and user name on any machine with perl.
Last I checked, this is never used.
The standard `file-attributes' does not have this `nonnumeric' argument
and the rcp.el code never uses it either.
* Very simple lisp for
"Pete" == Pete Forman [EMAIL PROTECTED] writes:
I did realise that but we are not going to be able to express all the
functionality in a short name. The most common use of the UNIX su
That's whay I suggested `shellfs' since it lays a filesystem on top of any
kind of shell access, whether
"Tom" == Tom Roche [EMAIL PROTECTED] writes:
How about FBT: "flying birch tree" :-) [EMAIL PROTECTED]
I'd vote for Kai Großjohann's baby (aka KGB), then.
Stefan
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
Is the complicated mechanism really necessary, or can we make do with
a fairly simple mechanism which caches information for a short time only?
I think that the sample you showed earlier indicates that we can already
win big-time with a
"Kai" == Kai Grossjohann [EMAIL PROTECTED] writes:
Don't use `when' rather than `if' when checking for EFS, as this would
require CL.
I don't know about XEmacs, but in Emacs, `when' has been standard at least
since Emacs-20. And since I believe that CL is dumped into XEmacs (or
at least, it
I see the above form in various places in rcp.el.
What is it for ? It seems not to do anything at all, except maybe for
silencing the byte-compiler. If so, I think a (defvar file) would
be preferable (i.e. more explicit).
Stefan "Not convinced that silencing the byte-compiler is
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
file-directory-p returns t for symlinks pointing to directories, but
`test -d' is false in such cases.
In what environment ?
`test -d' for me has always followed symlinks and it would feel very wrong
(contrary to the usual "when to follow
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
Ah. I thought that rcp-handle-file-directory-p calls `test -d', but,
no,
and why not ?
Stefan
PS: If there's a good reason, it should deserve a comment in the code
to make sure someone else won't go through the same problem.
"Pete" == Pete Forman [EMAIL PROTECTED] writes:
(defvar running-xemacs (string-match "XEmacs\\|Lucid" emacs-version))
(if running-xemacs
(require 'efs))
Parsing `emacs-version' is *evil* !
Why not do something like
(condition-case () (require 'efs) (error nil))
or
"Kai" == Kai Grossjohann [EMAIL PROTECTED] writes:
Implement `file-newer-than-file-p'. This implementation relies on the
`test' command to grok the `-nt' switch. Does this work on Irix, too?
It works on AIX.
The Irix machines I have access to are really badly setup, but
their `test' does
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
Did you (require 'rcp) before (require 'efs)? Or vice versa? You
should be doing (require 'efs) before (require 'rcp).
And jka-compr should be activated after all of those.
I think something should be done regarding the ordering on the
The two bugs below:
;; * Saving a file with execute permissions doesn't preserve the execute
;; permissions. (Reported by Dan McGuirk [EMAIL PROTECTED])
;; * Greg Stark: save a read-only file, Emacs asks whether to save
;; anyway, then tries to chmod the file, which fails.
are probably
Actually, there was a stupid bug in my code where set-file-modes
was returning the integer return status of the `chmod' rather
than throwing an error if the chmod failed.
In the mean time, I've implemented file-ownership-preserved-p and
cleaned up my current "new VC" patch a little.
Also, I've
[[ Mailed copy of news article posted to lists.emacs.rcp ]]
"David" == David Barr [EMAIL PROTECTED] writes:
Unfortunately, most of them don't have mimencode.
There's your problem, and you know how to fix it.
Stefan
[[ Mailed copy of news article posted to lists.emacs.rcp ]]
"Pete" == Pete Forman [EMAIL PROTECTED] writes:
a) Textually truncate the inode string before converting to int.
b) Set inode to 0 if error in read.
c) Maybe dired.c in newer versions of [X]Emacs has a workaround.
d)
[[ Mailed copy of news article posted to lists.emacs.rcp ]]
"Kai" == Kai Großjohann [EMAIL PROTECTED] writes:
Hm. That's bad. This means that it becomes rather more difficult to
use it. On the one hand, the syntax is the same for all uuencode
flavors, but on the other hand I have to
[[ Mailed copy of news article posted to lists.emacs.rcp ]]
"Kai" == Kai Grossjohann [EMAIL PROTECTED] writes:
I'm all for a simple solution, but I'm afraid it is not so simple.
Suppse that you use telnet to log in and the file contains the byte
sequence C-] q u i t RET? (Or you use
But ssh is not free, so rcp.el must be able to work with rsh (and this
must be the default) if it is to be distributed with Emacs, says Richard.
I'm not arguing this point. I'm arguing the usefulness of being able to use
*both* rsh and ssh at the same time via /s: and /r:
Stefan
"Kai" == Kai Grossjohann [EMAIL PROTECTED] writes:
Just `cat /file' doesn't seem to be right, somehow. Use `uuencode'
and `uudecode'? Is that available on all systems?
uuencoding seems overkill. It's probably enough to use an end-of-file marker
like "EOF" and to make sure it won't appear
[[ Mailed copy of news article posted to lists.emacs.rcp ]]
"Kai" == Kai Grossjohann [EMAIL PROTECTED] writes:
Greg Stark [EMAIL PROTECTED] writes:
The solution is either to ensure that the temporary files are
always created with the O_EXCL and O_CREAT flags set, or to create
a directory
61 matches
Mail list logo