the predictable MDTM anyway.
Readded -q to help cls.
Added missing \n to help cls (where'd that go?)
Added -1 to cls: single-column output. (This aligns with fileutils ls
-1.) This also cleans up -l.
--
Glenn Maynard
Index: FileSetOutput.cc
places
this code can be used.
nlist status output is still broken. The only fix I can see is the one
I had originally, so I didn't bother repeating it. :)
--
Glenn Maynard
Index: configure.in
===
RCS file: /home/lav/cvsroot/lftp
if
it was used for LIST); apparently the word-reading optimizations used
by memchr and memmove offset the penalty of iterating twice.
At 64 bytes/line for 10 megs of data, on my system, it's still 300ms.
The speed difference doesn't really matter.
--
Glenn Maynard
don't have access
to one of these to write an autoconf test, though.
--
Glenn Maynard
:
This one is nicer indeed. And smaller. It has a bug but it does not matter.
Er, what is it? Passed my tests (leading, trailing, consecutive \r\n
with or without null terminator.)
--
Glenn Maynard
.)
--
Glenn Maynard
file
/home/lav/cvsroot/CVSROOT/history
Permission denied
--
Glenn Maynard
(not sure
which.) Something else needs a fix, too, so the above works; probably one
of the parsers.
CVS is still down ...
--
Glenn Maynard
that it
might start happening in a release some day ...
--
Glenn Maynard
rejected: Access denied
I'm not going to jump through hoops to get mail sent to this host. If
you can send mail to this person, mind mentioning that it's bouncing
mail from my host?
--
Glenn Maynard
it?) Of course, if it's for compliance,
we should still send it when exiting, too.
--
Glenn Maynard
still finding Timer confusing? (Asking since you apparently
removed its use.)
--
Glenn Maynard
.
--
Glenn Maynard
, by multiplying
another interval by a scalar, etc; I wouldn't associate classes with single
operations. Avoiding TimeInt to avoid confusion with int.) The penalty
is a vptr; minor, since we don't carry around tons of times.
(Sorry for being a pest. :)
--
Glenn Maynard
readline-4.2/config.h.in missing.
--
Glenn Maynard
to 4. This affects nlist, find, du, cls, others.
By the way, do we really need to fall back on NLST after LIST when there
was no data? It's useful if we got data we couldn't parse, but if we
received nothing at all, there probably simply aren't any files.
--
Glenn Maynard
directory stack to count sizes.) I think I can simplify
FinderJob a bit; I'll get back to that once I try it.
--
Glenn Maynard
Index: FileSet.cc
===
RCS file: /home/lav/cvsroot/lftp/src/FileSet.cc,v
retrieving revision 1.19
diff
. On latent servers,
establishing lots of temporary data connections is a massive bottleneck ...
By the way, du is implemented; I'll wait until you get through the
previous small patch before sending it.
--
Glenn Maynard
On Thu, Nov 08, 2001 at 10:52:25AM +0300, Alexander V. Lukyanov wrote:
On Wed, Nov 07, 2001 at 12:47:50AM -0500, Glenn Maynard wrote:
readline-4.2/config.h.in missing.
It should be created by autogen.sh, which in turn calls autoheader.
It's blowing up:
02:59am [EMAIL PROTECTED]/6 [~/lftp
by
/usr/share/autoconf2.13/acconfig.h
I use autoconf 2.52.
So do I (well, 2.50); readline has an old AC_PREREQ, which makes it fall
back on 2.13.
--
Glenn Maynard
. This one's borderline. (I can probably implement this myself;
I'll get back to it later.)
--
Glenn Maynard
On Thu, Nov 08, 2001 at 03:38:11AM -0500, Glenn Maynard wrote:
I don't know. That might be useful, though I'm not sure what it would
do over a regular buffered FDStream. I don't think it'd be needed for
this ...
A thought: if subshells set up a pty, we could still output the status bar
when
locally or remotely ... we do have
the completion code that determines this. Then we wouldn't have to
worry about it at all at the command level.
du later tonight.
--
Glenn Maynard
On Thu, Nov 08, 2001 at 12:55:05PM +0300, Alexander V. Lukyanov wrote:
Install gettext-0.10.40 - then patches should apply cleanly.
Perhaps autogen.sh should check the version?
--
Glenn Maynard
patch fixes it.
Well, the problem is there are other 550s which are actually errors;
550 Access denied. should go to 0 (almost all should, actually);
only 550 No files found. should go to 4.
--
Glenn Maynard
. :)
--
Glenn Maynard
On Thu, Nov 08, 2001 at 10:46:30PM -0500, Glenn Maynard wrote:
I don't think ListInfo should handle path-to-file argument. Anyway, it
would have to list all parent directory to get the info.
It would have to list the parent to do this, yes (another reason for it
to be optional
validation the SSL layer
gives, which is no less important anonymously.
--
Glenn Maynard
On Thu, Nov 08, 2001 at 10:46:30PM -0500, Glenn Maynard wrote:
The only tricky part: which directory is the FileSet actually in? Two
options: we can make the class itself make this available, or we can add
a char *dir to FileSet. Any reason not to do the latter? It'd be nice
to keep
. The only reason to use
ls instead of cls is because ls outputs data as it's received; cls has
to wait for all data to finish before it can output. It's not possible
to merge them, however.
If you like cls, use
alias ls cls
alias cls ls
in .lftprc.
--
Glenn Maynard
work out these issues, since
it's not useful yet. (I've attached the class source, if you want to look
at it; don't drop it in yet, though.)
The message was without an attachement.
Duh. Sorry.
--
Glenn Maynard
#include config.h
#include errno.h
#include GetFileInfo.h
#include misc.h
of
FinderJob's listing and CDing, but that's no good: we'd end up with full
paths (relative to the root of the find, anyway), which would use a ton
of memory for large trees.
--
Glenn Maynard
Index: FileSet.cc
===
RCS file: /home/lav
, it should
probably only live for a single operation. ls doesn't do this since
it's implicitely single-op.)
--
Glenn Maynard
Index: FileSet.cc
===
RCS file: /home/lav/cvsroot/lftp/src/FileSet.cc,v
retrieving revision 1.22
diff -u
, in fact (try put completion in /usr/bin), so
I'll try to get this implemented today. (LsCache object, intended scope
one operation, members getpwuid, getgrgid, maybe getpwnam/getgrnam.)
--
Glenn Maynard
.
--
Glenn Maynard
.
I've asked for this before, but we didn't get a consensus.
Alex, want this? It won't hurt anything.
--
Glenn Maynard
is that the cache be flushed,
minimally, between every two interactive (or repeat, etc.) commands.
Implemented ftp:ssl-allow-anonymous.
Added closure on hostname for ssl options.
By the way, I like the mirror ordering--SFV files have always been good
at being downloaded last. :)
--
Glenn Maynard
);
textdomain (PACKAGE);
I can't reproduce this with LC_ALL=pl_PL.ISO-8859-2; what's needed?
--
Glenn Maynard
that. This doesn't make it impossible to merge m*
commands, though.
--
Glenn Maynard
() (isscanf?) A wrapper should
suffice; save locale, set, sscanf, restore ...
--
Glenn Maynard
does its own globbing; it never sends a literal wildcard to the
server (unless it's part of a filename or something like literal
arguments to ls.)
Set debug to something high (5 or so) and do an mget and you'll see
what's going on.
--
Glenn Maynard
cache that. We could add a dummy cache entry with an unused type,
like FA::CHANGE_DIR, and no data, if this was needed.)
--
Glenn Maynard
# Directly out of glib. We don't do copy-by-value and other really
# pessimistic tests, since just about all systems will have one of these.
# Add them
On Thu, Nov 22, 2001 at 02:57:29PM -0500, Glenn Maynard wrote:
glob -a mget *blah*
and it'll have this behavior. -a tells it to get *all* files, which
makes it use NLST and have the above partially recursive
behavior--behavior which I consider broken, incidentally (when did I ask
On Fri, Nov 23, 2001 at 03:28:32PM -0500, Glenn Maynard wrote:
with those other places corrected. Completion should be corrected for
case insensitivity too.
That's easy enough. I'll send a patch once you apply what you're going
to apply, though, so we don't overlap.
I'll just send
missing.
--
Glenn Maynard
in a wide screen--worst case--was 100ms to 40ms or so for that call on
this system; that's not a bad gain (would matter on slow systems), but
it complicates the function too much. I'm mostly posting it to make it
available, since I'm rolling back my ColumnOutput.cc.
--
Glenn Maynard
Index
, we should never retry if the error means not a
directory; in practice we're probably better off never persisting at
all here, since this particular special case is important--we can't have
it failing on oddly worded or multilingual servers.)
--
Glenn Maynard
, without slowing
things down. It should maintain the current jobs -v behavior, though
(it's useful) ...
It does *almost* work: if I'm careful and match directories, I can queue
a user command and it'll work.
--
Glenn Maynard
been updated, obviously ...
--
Glenn Maynard
having multiple
uses, and I don't think it needs to be fixed. (Yeah, you could prevent
that with two classes, but it doesn't gain you anything and you end up
with a lot more code.)
By the way, I'm going to clean up the Timer class interface a bit; patch
probably inbound later today.
--
Glenn Maynard
On Tue, Oct 09, 2001 at 12:21:03PM -0400, Glenn Maynard wrote:
It was a little more direct, originally, storing the interval with its
constructor; I removed that as the interval can change with each call.
I might put it back and update the interval with Reconfig, which would
make it more
needed? Locales convert to different encodings automatically.
--
Glenn Maynard
Index: .cvsignore
===
RCS file: /home/lav/cvsroot/lftp/.cvsignore,v
retrieving revision 1.5
diff -u -r1.5 .cvsignore
--- .cvsignore 2001/03/20 09:28:46
not send them,
so it does not matter for client.
Hmm, wonder why it was doing that, then.
--
Glenn Maynard
FTP servers.)
--
Glenn Maynard
Index: CmdExec.cc
===
RCS file: /home/lav/cvsroot/lftp/src/CmdExec.cc,v
retrieving revision 1.82
diff -u -r1.82 CmdExec.cc
--- CmdExec.cc 2001/10/05 06:26:49 1.82
+++ CmdExec.cc 2001/10/17 01:47
On Thu, Nov 29, 2001 at 01:58:56PM -0500, Glenn Maynard wrote:
It would be good to have ftp:disallow-chmod or some option in mirror
command...
I'll add ftp:use-site-chmod setting.
I wouldn't recommend that--not for this reason alone, anyway. Note that
this is exactly the same
Suggestion: make mirror:order case-insensitive. I've seen lots of all-caps
SFV files, etc.
On Sun, Dec 02, 2001 at 03:36:04AM -0500, Glenn Maynard wrote:
GetFileInfo.cc: Propagate use_cache to child ListInfos; fixes recls.
Don't EACCESS if we don't know the file type.
I've been bashing
But they are very powerfull, and it's fun writing them :)
I'm definitely not suggesting ont supporting regexes; they're useful.
--
Glenn Maynard
it always uses -dils.
(It'd be easy enough to implement, as long as we force -l, but ls -R is
far more useful, I think.)
--
Glenn Maynard
On Fri, Dec 14, 2001 at 03:06:04PM -0500, Glenn Maynard wrote:
Is there any way to get these particular headers to define mbstate_t
properly in C++? It'd be nice to avoid the extra complication of
needing to test for mbstate_t differently in C and C++.
Er, I guess that could be hidden inside
.
--
Glenn Maynard
Index: Http.h
===
RCS file: /home/lav/cvsroot/lftp/src/Http.h,v
retrieving revision 1.36
diff -u -r1.36 Http.h
--- Http.h 2001/12/05 15:03:43 1.36
+++ Http.h 2001/12/15 03:09:05
@@ -132,6 +132,7
with closure on protocol would be useful.)
--
Glenn Maynard
On Sat, Dec 15, 2001 at 08:58:12PM -0500, Glenn Maynard wrote:
In FileCopyPeerCLS, you set can_seek/can_seek0 to false. Is this
needed? It's a buffer, so it should be able to seek if necessary. (It
may become useful in the future to have OutputJob seekable, too.)
Okay, adding to buffers
edits fixed this. I'll update the top comment
with this change, too.
--
Glenn Maynard
any clever
logic, so they should be much more straightforward. :) (But they can
wait until this is stabilized.)
--
Glenn Maynard
prematurely. Also,
printf(%i %i\n, j-Done(), j-Done()); was doing 1 0. One would
close and return IN_PROGRESS, the next would return OK.
--
Glenn Maynard
On Tue, Dec 18, 2001 at 04:01:12AM -0500, Glenn Maynard wrote:
I've added a number of cases at the top of the file, and tested each
case individually. Currently, all of them work reasonably; one is
pulling the error message form the wrong command. Once I fix that and
redo the tests, I'll
going to try to get them to
change LIST to return errors sanely ... but we're always stuck with
legacy servers, of course.)
--
Glenn Maynard
Index: FileAccess.cc
===
RCS file: /home/lav/cvsroot/lftp/src/FileAccess.cc,v
retrieving
, it was lftp sending this header, my mistake.
Yep. Client sends it to indicate support; server responds to OK it, I
believe.
--
Glenn Maynard
in them.
queue get sth1
queue get sth2
won't notify you when sth1 is downloaded.
Well, maybe we should have a report-transfer-completion {yes,no,fg},
the latter being the current behavior (which I would prefer to be the
default.)
--
Glenn Maynard
just cosmetic.
--
Glenn Maynard
On Sun, Dec 23, 2001 at 10:57:16PM -0500, Glenn Maynard wrote:
Mind repeating each of these, doing a QUOTE PWD after each? I want to
know if the paths output are *actually* where it's ending up, or if
that's just cosmetic.
Oh, and exactly what server version is this happening with? New CVS
, then LC_CTYPE will have effect.
Ah. He should be using LANG, not LC_ALL.
--
Glenn Maynard
pathESC[... A better fix might be to 1: find the
first and last quotes; 2: scan-copy them, doing - ; if we ever hit a
not followed by a , then 3: redo 2, not doing - .
--
Glenn Maynard
Index: ftpclass.cc
===
RCS file: /home/lav
by the session with highest priority.
While I don't agree with the logic, this doesn't particularly matter right
now. The only possible use I could ever see for this is for an antiidle
task, and that would want to wait, too (for a different reason), so I don't
want this removed anyway.
--
Glenn
... Yes I know it's a bit crazy :)
That's fine (though I would argue it's a useless goal); the problem is
that the output isn't matching what should be happening.
--
Glenn Maynard
files linked
from someweher else protection, however, this also makes bookmarks a bit
useless and doesn't work over many proxies (privacy), so these servers are
fortunately a dying thing.
Unfortunately, they seem as prevelent as ever. Geocities does this for
all images ...
--
Glenn Maynard
confuse the referer with something else after all?
I've definitely seen these pages giving not allowed type stuff for
images. Recently, people have been circumventing this by renaming
images to *.txt (abusing an IE bug), so they probably just dropped it.
--
Glenn Maynard
things like close up FDs. This leads to some hard-to-debug
problems, because this is a rather obscure side effect.)
--
Glenn Maynard
On Fri, Jan 04, 2002 at 06:50:02PM +0300, Alexander V. Lukyanov wrote:
Thanks, I have fixed it by checking delete_second. zcat for multiple
files works too now.
My version of CatJob (using OutputJob) started working, too.
--
Glenn Maynard
, could you also include a email reply? i've not subscribed
to the mailing list.
Tip: when you want this, add a header
Mail-Copies-To: Lethal Weapon [EMAIL PROTECTED]
to your post; it'll hint most decent MUA's to do just that.
--
Glenn Maynard
a solution (it'd end up
killing the fg process, not the one running the filter ...)
--
Glenn Maynard
somewhere if you want to try it. (The intention
is to call that from the parser.)
--
Glenn Maynard
Index: ChmodJob.cc
===
RCS file: /home/lav/cvsroot/lftp/src/ChmodJob.cc,v
retrieving revision 1.5
diff -u -r1.5 ChmodJob.cc
On Mon, Jan 21, 2002 at 11:02:35AM +0100, Piotr Krukowiecki wrote:
Next segfault, the same lftp version (20020113)
Can you reproduce this?
--
Glenn Maynard
do things
like command | pipe ftp://url;. (Adding URL redirection but saying
... but you can't combine it with pipes wouldn't be very good.)
--
Glenn Maynard
.
(Actually, I think I've found a way to reproduce it.) Readline is
exiting with errno == EINTR--so the readline package is compiled wrong.
(That's what I've been suspecting.) I'll bring this up with them.
--
Glenn Maynard
):
(Actually, the above is perfectly correct behavior from lftp; wingate is
broken.)
Use set ftp:ssl-allow no.
--
Glenn Maynard
thinking of something like an OutputFilter derivation to do just
for_each handling.
Anyway, I'd like to make another major release before integrating that.
I agree, there have been a lot of small fixes.
--
Glenn Maynard
with it. Being rid of the
entire reattach child processes deal would be a lot simpler.
--
Glenn Maynard
, and this
would mean I'd have to change habits for these function calls.)
--
Glenn Maynard
On Tue, Jan 22, 2002 at 12:21:26AM -0500, Glenn Maynard wrote:
On Mon, Jan 21, 2002 at 11:05:33PM -0500, Glenn Maynard wrote:
And I just hit the weird lftp quitting spontaneously thing again.
(Actually, I think I've found a way to reproduce it.) Readline is
exiting with errno == EINTR--so
to fix
it up, but I probably won't get to it soon.
--
Glenn Maynard
.
--
Glenn Maynard
---BeginMessage---
On Thu, 28 Feb 2002 13:26:15 -0500
Glenn Maynard [EMAIL PROTECTED] wrote:
I think binding to an address is important; if the last is implemented
(interface name) I'd strongly suggest it be secondary to IP. (I don't
think interfaces names could be portable
On Sun, Mar 03, 2002 at 09:06:13PM +0300, Alexander V Lukyanov wrote:
It is orbZ, not orbs Any opinions on making lftp-devel SPAMFRIEND?
(I have an allergy to final Zs) Don't know what that is; looks like a
sendmail thing, and I never touch sendmail
--
Glenn Maynard
or things may not work right is
reasonable, IMO
--
Glenn Maynard
be (Alex could answer that)
(Note that this isn't necessary to allow hostnames in IP's for that
setting; it would just be a restriction)
--
Glenn Maynard
0x080890bf in FileAccess::Chdir (this=0x8113c88, path=0x8108880 ., verify=true)
at FileAccess.cc:690
690char *newcwd=(char*)alloca(strlen(cwd)+strlen(path)+2);
(gdb) print cwd
$1 = 0x0
--
Glenn Maynard
select
the mirror job by number.
--
Glenn Maynard
there is a limitation: a job can be awaited by a single other job.
When you do fg, you make the main CmdExec parent of the selected job, which
is not possible when some other job is already parent of the job.
That's what I meant: fg the queue.
--
Glenn Maynard
with multiple programs and servers.
--
Glenn Maynard
On Tue, Apr 02, 2002 at 10:48:25AM -0500, Justin Piszcz wrote:
Downloader for X does, etc, but I need a console tool thats as reliable as
wget/lftp.
He was suggesting to try them to see if rollback helps, not to use them as a
permanent replacement.
--
Glenn Maynard
1 - 100 of 158 matches
Mail list logo
