filename contains ":" characters in scp?
Raymond Chui
myfile2001-05-15_09:30:00.txt
when I do scp command, I try
scp myfile2001-05-15_09\:30\:00.txt userid@remote_host:dir_path/
But the scp command in ssh 2.3 (in Redhat 6.2) doesn't recognized
the "\" character (back slash). I need "\" because file name contains
":"
which is part of scp command. In UNIX character "\" should treat next
character as part of filename (in this case is ":" colon).
Shall ssh 2.5 solved this bug? Thank you!
--Raymond
Lew Pitcher
<raymon...@noaa.gov> wrote:
>This is a multi-part message in MIME format.
>--------------9283BCDF072D24F0774570A6
>Content-Type: text/plain; charset=us-ascii
>Content-Transfer-Encoding: 7bit
>
>I have some files their name contain the ":" colon character like
>
>myfile2001-05-15_09:30:00.txt
Which is a legal filename, of course.
>when I do scp command, I try
>
>scp myfile2001-05-15_09\:30\:00.txt userid@remote_host:dir_path/
Why? Colons in filenames are legal in the shell.
>But the scp command in ssh 2.3 (in Redhat 6.2) doesn't recognized
>the "\" character (back slash). I need "\" because file name contains
>":"
>which is part of scp command. In UNIX character "\" should treat next
>character as part of filename (in this case is ":" colon).
But the colon character is carried properly in filename; escaping it
in the shell won't change the filename any. You can't fix scp's
problem with the filename by escaping the colon in the shell. All
you've done is instruct the shell to pass the colon as part of the
filename, and that seems to be something that scp doesn't like. You'll
have to fix or work-around scp.
A suggestion: since scp has problems with files that include colons in
the filename, change your filenames to exclude the colons.
>Shall ssh 2.5 solved this bug? Thank you!
>
>--Raymond
>
>
>--------------9283BCDF072D24F0774570A6
>Content-Type: text/x-vcard; charset=us-ascii;
> name="raymond.chui.vcf"
>Content-Transfer-Encoding: 7bit
>Content-Description: Card for Raymond Chui
>Content-Disposition: attachment;
> filename="raymond.chui.vcf"
>
>begin:vcard
>n:Chui;Raymond
>tel;fax:(301)713-0963
>tel;work:(301)713-0624 Ext. 168
>x-mozilla-html:TRUE
>url:http://members.xoom.com/rchui/
>org:NWS, NOAA
>version:2.1
>email;internet:Raymon...@noaa.gov
>title:SA, DBA
>note:ICQ #: 16722494
>adr;quoted-printable:;;NOAA, NWS, Office of Hydrology, OH=0D=0A1325 East-West Highway, Room 8112;Silver Spring;MD;20910-3283;U.S.A.
>x-mozilla-cpt:;-6384
>fn:Raymond Chui
>end:vcard
>
>--------------9283BCDF072D24F0774570A6--
>
Lew Pitcher, Information Technology Consultant, Toronto Dominion Bank Financial Group
(Lew_P...@td.com)
(Opinions expressed are my own, not my employer's.)
Peter T. Breuer
> I have some files their name contain the ":" colon character like
> myfile2001-05-15_09:30:00.txt
> when I do scp command, I try
> scp myfile2001-05-15_09\:30\:00.txt userid@remote_host:dir_path/
That's silly! There is no need to escape the ":". You are escaping it
to the SHELL, not to the scp.
> But the scp command in ssh 2.3 (in Redhat 6.2) doesn't recognized
> the "\" character (back slash). I need "\" because file name contains
I am FED UP explaining to people who ought to know better it that the
programs does not see your "escape". The shell does! What is so hard to
understand about this?
> ":"
> which is part of scp command. In UNIX character "\" should treat next
> character as part of filename (in this case is ":" colon).
And it DOES (except that you mean "argument", not "filename").
> Shall ssh 2.5 solved this bug? Thank you!
It is not a bug.
Don your thinking apparatus and realize the solution to your
non-problem.
Peter
KW
separators for hosts. The only solution I can think of is to copy the
files to a separate folder and scp the entire folder recursively...
In article <3B0116A...@noaa.gov>, "Raymond Chui"
Erik Max Francis
> Shall ssh 2.5 solved this bug? Thank you!
It's not a bug. scp sees colons as separating the hostname from the
remote filepath. scp doesn't play well with colons in filenames.
(Escaping the colons doesn't do anything; the shell isn't what's having
a problem with them, it's scp.)
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ No one knows what he can do until he tries.
\__/ Publilius Syrus
Product's Quake III Arena Tips / http://www.bosskey.net/
Tips and tricks from the absolute beginner to the Arena Master.
Raymond Chui
>
>
> > when I do scp command, I try
>
> > scp myfile2001-05-15_09\:30\:00.txt userid@remote_host:dir_path/
>
> That's silly! There is no need to escape the ":". You are escaping it
> to the SHELL, not to the scp.
>
>
>
> It is not a bug.
>
Then why
cp myfile2001-05-15_09:30:00.txt dir_path/
work OK, but
scp myfile2001-05-15_09:30:00.txt userid@remotehost:dir_path
not work?
because scp command sees the ":" character as separator for host name
and port number. If filename contains ":", it thinks as port number is 30 in
above case.
That is a problem in scp command!!
--Raymond
Erik Max Francis
> Then why
>
> cp myfile2001-05-15_09:30:00.txt dir_path/
>
> work OK, but
>
> scp myfile2001-05-15_09:30:00.txt userid@remotehost:dir_path
>
> not work?
>
> because scp command sees the ":" character as separator for host name
> and port number.
No, it separates the host name and remote path. You specify a
non-standard port with the -P option.
> If filename contains ":", it thinks as port number is 30 in
> above case.
No, it is the way it is supposed to work. Don't try to scp files that
contain colons.
--
Erik Max Francis / m...@alcyone.com / http://www.alcyone.com/max/
__ San Jose, CA, US / 37 20 N 121 53 W / ICQ16063900 / &tSftDotIotE
/ \ There is no fate that cannot be surmounted by scorn.
\__/ Albert Camus
Official Buh rules /
http://www.alcyone.com/max/projects/cards/buh.html
The official rules to the betting card game, Buh.
Bill Unruh
Arrogant bastard aren't you.
He asks a valid question ( and yes does not understand shell escaping vs
program escaping) and you vent your spleen. Please go have another cup
of coffee.
...
]That's silly! There is no need to escape the ":". You are escaping it
]to the SHELL, not to the scp.
]I am FED UP explaining to people who ought to know better it that the
]programs does not see your "escape". The shell does! What is so hard to
]understand about this?
Then stop explaining and stop posting.
...
]It is not a bug.
Bill Unruh
]Raymond Chui wrote:
]> Shall ssh 2.5 solved this bug? Thank you!
]It's not a bug. scp sees colons as separating the hostname from the
]remote filepath. scp doesn't play well with colons in filenames.
](Escaping the colons doesn't do anything; the shell isn't what's having
]a problem with them, it's scp.)
Yes, and he is asking, is there a way to get scp to accept colons in
filenames. Apparently none of you know. I do not either.
And yes, it is a bug if scp gives no way of escaping the colon within a
valid filename. (neither the scp or rcp man pages give any hint, so it
may well be a bug in scp)
Peter T. Breuer
>> > when I do scp command, I try
>>
>> > scp myfile2001-05-15_09\:30\:00.txt userid@remote_host:dir_path/
>>
>> That's silly! There is no need to escape the ":". You are escaping it
>> to the SHELL, not to the scp.
>> It is not a bug.
> Then why
> cp myfile2001-05-15_09:30:00.txt dir_path/
> work OK, but
It does NOT work OK.
oboe:/usr/oboe/ptb% scp myfile2001-05-15_09:30:00.txt /tmp
Bad host name: myfile2001-05-15_09
> scp myfile2001-05-15_09:30:00.txt userid@remotehost:dir_path
> not work?
It DOES work. It does exactly what it is supposed to do. Actually, if
you want to know, your example does not work:
oboe:/usr/oboe/ptb% scp myfile2001-05-15_09:30:00.txt ptb@localhost:dir_pat
Bad host name: myfile2001-05-15_09
> because scp command sees the ":" character as separator for host name
That's because it IS.
> and port number. If filename contains ":", it thinks as port number is 30 in
No, no port number. You are imagining that. The syntax is:
scp [-aAqQprvBCL1] [-S path-to-ssh] [-o ssh-options]
[-P port] [-c cipher] [-i identity]
[[user@]host1:]filename1... [[user@]host2:]filename2
> above case.
> That is a problem in scp command!!
No, it is a problem in your head. Please sort it out. Get used to it.
Computers do what you tell them to. The first ":" in an argument to
scp is interpreted as the separator between a hostname and a path.
That's documented behaviour. Be grateful that the first part separated
thusly in your second example is not a valid hostname.
Peter
Bill Unruh
>> If filename contains ":", it thinks as port number is 30 in
>> above case.
>No, it is the way it is supposed to work. Don't try to scp files that
>contain colons.
No, it is a bug. If it regards : as a special character, it should also
give a way of escaping that special character. If it does not, then that
is a bug. It is not"the way it is supposed to work". It was just
carelessness on the part of the writers, who I suspect would agree.
: is a valid filename character and as such should be useable in a
filename to scp.
Matt D
> > Then why
>
> > cp myfile2001-05-15_09:30:00.txt dir_path/
> > work OK, but
cp ie NOT scp.
Matt
Bob Hauck
>No, it is a bug. If it regards : as a special character, it should also
>give a way of escaping that special character.
It is only the first ':' that is special. He's copying between local files,
so...
[hauck@lab hauck]$ echo "boo" > file:with:colons
[hauck@lab hauck]$ scp localhost:file:with:colons /tmp
file:with:colons 100%|*****************************| 4 00:00
[hauck@lab hauck]$ ls /tmp/file:with:colons
/tmp/file:with:colons
Now, that wasn't so hard, was it?
--
-| Bob Hauck
-| Codem Systems, Inc.
-| http://www.codem.com/
rdh
<p...@oboe.it.uc3m.es> wrote:
>In comp.os.linux.help Raymond Chui <raymon...@noaa.gov> wrote:
>> I have some files their name contain the ":" colon character like
>
>> myfile2001-05-15_09:30:00.txt
>
>> when I do scp command, I try
>
>> scp myfile2001-05-15_09\:30\:00.txt userid@remote_host:dir_path/
>
>That's silly! There is no need to escape the ":". You are escaping it
>to the SHELL, not to the scp.
>
>> But the scp command in ssh 2.3 (in Redhat 6.2) doesn't recognized
>> the "\" character (back slash). I need "\" because file name contains
>
>I am FED UP explaining to people who ought to know better it that the
>programs does not see your "escape". The shell does! What is so hard to
>understand about this?
Nothing, if a person has been around Unix environments as long as you
have, Peter. However, I've noticed that you tend to assume that even
the most green newbies should have an understanding of these concepts
as well. Reminds me of several university professors that I've had
the misfortune of studying under.
"But... but... but... this is simply an application of
$OBSCURE_PRINCIPLE_OF_QUANTUM_PHYSICS! You should have known that
from birth!"
--
Russell rdh at salug dot org
"I don't like the way typical females act. Neither do I like
the way typical males act. I loathe the vast majority of
humanity." --mdxi, K5
Peter T. Breuer
> In <6n9rd9...@news.it.uc3m.es> "Peter T. Breuer" <p...@oboe.it.uc3m.es> writes:
> Arrogant bastard aren't you.
> He asks a valid question ( and yes does not understand shell escaping vs
It is not a valid question. If you think that, then "why do you always
put on pink trousers every wednesday? Is it a bug?" is a valid question
of mine :-)
Peter
Peter T. Breuer
> On Tue, 15 May 2001 15:07:50 +0200, "Peter T. Breuer"
> $OBSCURE_PRINCIPLE_OF_QUANTUM_PHYSICS! You should have known that
> from birth!"
So they should. Quantum physics has been around since the 1920s. I
certainly learned basic quantum mechanics (hydrogen atom, etc.)
before I went to university. Mind you, on the "big brother" TV series
here, I saw that they took the people aside to give them some tests,
and several of them could neither do long division, nor evaluate
numbers given in roman numerals. And I recall talking to a biology
major in the states and finding that she didn't know what a myelin
sheath was. So apparently "common knowledge" is not common to many
people.
Peter
Peter T. Breuer
> Yes, and he is asking, is there a way to get scp to accept colons in
Then let him Ask That, instead of harping on about nonexistant bugs.
> filenames. Apparently none of you know. I do not either.
No all of us "know", because it is obvious.
> And yes, it is a bug if scp gives no way of escaping the colon within a
> valid filename. (neither the scp or rcp man pages give any hint, so it
> may well be a bug in scp)
They don't have to give a hint, when it's specified right up there in
the first line of the manpage. For your amusement:
scp [-aAqQprvBCL1] [-S path-to-ssh] [-o ssh-options]
[-P port] [-c cipher] [-i identity]
[[user@]host1:]filename1... [[user@]host2:]filename2
See it?
Peter
Peter T. Breuer
>>> If filename contains ":", it thinks as port number is 30 in
>>> above case.
>>No, it is the way it is supposed to work. Don't try to scp files that
>>contain colons.
> No, it is a bug. If it regards : as a special character, it should also
> give a way of escaping that special character. If it does not, then that
It DOES. That you seem incapable of understanding that is your problem!
Peter
Lew Pitcher
Hint for the clueless, one of the names for the local system is
'localhost'
Colin Watson
>In <6n9rd9...@news.it.uc3m.es> "Peter T. Breuer"
><p...@oboe.it.uc3m.es> writes:
>
>Arrogant bastard aren't you.
>He asks a valid question ( and yes does not understand shell escaping vs
>program escaping) and you vent your spleen. Please go have another cup
>of coffee.
>...
>]That's silly! There is no need to escape the ":". You are escaping it
>]to the SHELL, not to the scp.
>
>
>]I am FED UP explaining to people who ought to know better it that the
>]programs does not see your "escape". The shell does! What is so hard to
>]understand about this?
>
>Then stop explaining and stop posting.
Believe me, his bug reports are no more useful. When half a dozen
different complaints are raised in the one long rant of a report (e.g.
Debian bug #47000), the inclination is to fix something else instead.
'tar cf - ... | ssh remote-host tar xvf -' can be a useful way to get
around scp's (apparent) lack of an escape.
--
Colin Watson [cj...@flatline.org.uk]
"Usenet is like a herd of performing elephants with diarrhea -- massive,
difficult to redirect, awe-inspiring, entertaining, and a source of mind-
boggling amounts of excrement when you least expect it." - "spaf", 1992
Dan Mercer
Sheesh - doesn't know what a myelin sheath is! It's where you keep
your myelin Bowie knife!
I don't have scp, but rcp (which has the same problem) works
fine if you give it the full path name:
rcp $PWD/file:withcolons dest:destdir
--
Dan Mercer
dame...@mmm.com
So apparently "common knowledge" is not common to many
> people.
>
> Peter
Opinions expressed herein are my own and may not represent those of my employer.
Peter T. Breuer
> On Tue, 15 May 2001 18:53:29 +0200, "Peter T. Breuer"
>> [-P port] [-c cipher] [-i identity]
>> [[user@]host1:]filename1... [[user@]host2:]filename2
>>
>>See it?
> Hint for the clueless, one of the names for the local system is
> 'localhost'
awww ... a puzzle spoiler! Should have been written upside down at the
bottom of the next page :-).
Peter
rdh
<p...@oboe.it.uc3m.es> wrote:
> So apparently "common knowledge" is not common to many
>people.
That's my point. Things that are "common knowledge" to UNIX admins
aren't necessarily common knowledge to everyone else. If one has only
heard of Linux within the last week, and only installed it yesterday,
then one isn't going to know as much as Peter T. Breuer.
These groups are full of new users, Peter. While your advice might be
beneficial to me and to several other regulars, the majority of Linux
newbies posting and lurking here might as well be sitting in a class
with my old professors. There is nothing more frustrating than being
a freshman undergraduate in a 101- or 102-level class, listening to a
professor who seems to think he's delivering a refresher course to his
fellow Ph.D.'s. That is the situation you put many of these newbies
in.
Don't get me wrong. I've been hanging around c.o.l.* for a while, and
you're one of the most knowledgable people I've seen. You've given me
numerous helpful tips (or given tips to others, which I read and found
applicable). But many of your responses to newbies are less than
helpful -- in fact, some of them go so far over the querent's head
that the poor soul's only options are to become utterly confused, or
to ignore your post entirely (an option made easier by the abrasive
manner in which some of your answers are delivered).
All I'm saying is that you might consider giving some of these people
a break. They want to learn, but that takes time. They don't yet
have the underlying concepts -- these *cannot* be glossed over when
trying to teach them. You just can't jump straight to the advanced
stuff. It will not work.
Sven Bovin
Yep, but you're probably one of the lucky bastards with an
LCD. It's such a PITA to turn my CRT upside down and look
in the backside :-)
Sven
--
-----------------------------------------------------------
sven dot bovin at chem dot kuleuven dot ac dot be
-----------------------------------------------------------
Wayne Pollock
either teach a simplified physics course that the majority can
understand and feel good about, or teach a tough course that only
majors will appreciate. (In my experience about 1 in 30 students.)
The simple approach is useless to those who plan a career in physics
and already understand the basics. (Notice how I avoid contention
by using physics not linux? :-) The expert approach leads to
student frustation and complaints.
The better schools have two courses, physics for majors and physics
for non-majors. But if your institution only has one course and
you must teach it, then you must choose which audience to teach
for. There is no approach that will satisfy all students.
In the early days of the Internet where there were only a few
dozen netnews groups, there were two unix groups only, "unix-wizards"
and another (the name escapes me). The experts with no patience
for newbie questions read and posted in the wizards group, those
who found newbie questions interesting (or amusing) used the
other. The system seemed to work, maybe its time for
"comp.os.linux.experts-only"?
Just my 2 cents, worth about $0.00000001 on the open market.
-Wayne Pollock
Wayne Pollock
[Author Unknown]
Name Description and features:
beginner - insecure with the concept of a terminal
- has yet to learn the basics of vi
- has not figured out how to get a directory
- still has trouble with typing
after each line of input
novice - knows that "ls" will produce a directory
- uses the editor, but calls it "vye"
- has heard of "C" but never used it
- has had his first bad experience with rm
- is wondering how to read his mail
- is wondering why the person next to him
seems to like Unix so very much.
user - uses vi and nroff, but inexpertly
- has heard of regular-expr.s but never seen one.
- has figured out that "-" precedes options
- has attempted to write a C program and has
decided to stick with pascal
- is wondering how to move a directory
- thinks that dbx is a brand of stereo component
- knows how to read his mail and is wondering
how to read the news
knowlegable - uses nroff with no trouble, and is beginning
user to learn tbl and eqn
- uses grep to search for fixed strings
- has figured out that mv(1) will move directories
- has learned that "help" doesn`t help
- somebody has shown him how to write C programs
- once used sed to do some text substitution
- has seen dbx used but does not use it himself
- thinks that make is only for wimps
expert - uses sed when necessary
- uses macro"s in vi, uses ex when neccesary
- posts news at every possible opportunity
- write csh scripts occasionally
- write C programs using vi and compiles with cc
- has figured out what "&&" and "||" are for
- thinks that human history started with "!h"
hacker - uses sed and awk with comfort
- uses undocumented features of vi
- write C code with "cat >" and compiles with "!cc"
- uses adb because he doesn`t trust source
debuggers
- can answer questions about the user environment
- writes his own nroff macros to supplement std.
ones
- write scripts for Bourne shell (/bin/sh)
- knows how to install bug fixes
guru - uses m4 and lex with comfort
- writes assembly code with "cat >"
- uses adb on the kernel while system is loaded
- customizes utilities by patching the source
- reads device driver source with breakfast
- can answer any unix question after a
little thought
- uses make for anything that requires two or more
distinct commands to achieve
- has learned how to breach security but no longer
needs to try
wizard - writes device drivers with "cat >"
- fixes bugs by patching the binaries
- can answer any question before you ask
- writes his own troff macro packages
- is on first-name basis with Dennis, Bill, and Ken
--------------------------------------------------------------------------------
Found on my hard disk, its 3 AM, so I can't tell if its
funny and/or appropriate.
-Wayne Pollock
Peter T. Breuer
> On the other hand, your physix <g> professor must make a choice:
> either teach a simplified physics course that the majority can
> understand and feel good about, or teach a tough course that only
> majors will appreciate. (In my experience about 1 in 30 students.)
It's not a question of "appreciation", but of "usefulness". There's no
point in teaching physics to people who can't understand physics, just
as there's no point in trying to teach me how to play golf (I'm not
interested, and probably wouldn't be any good at it anyway). As to
only majors? Physics dies not strike me as a subject that one can take
half-heartedly! It's like a side-plate of steak.
> The simple approach is useless to those who plan a career in physics
Yep.
> and already understand the basics. (Notice how I avoid contention
> by using physics not linux? :-) The expert approach leads to
> student frustation and complaints.
The solution is usually to split the class into talented students
and not-so-talented, and get very different people to teach them!
> The better schools have two courses, physics for majors and physics
> for non-majors. But if your institution only has one course and
Yes, just so.
> you must teach it, then you must choose which audience to teach
> for. There is no approach that will satisfy all students.
And also there is no single approach that works for one group of
students. My old prof of algebraic representation theory of groups
springs to mind .. he did everything backwards, and I loved it! He
first stated the theorem, then the lemma needed to prove that, then the
one needed to prove that, and so on, backwards. It was wonderfully easy
to follow. And all his results were about 5 lines long, so each one
was just a simple brainteaser, with no strategic planning requirement.
Most people hated it, as they didn't know which way they were going!
But I had no trouble. I much preferred that to huge proofs in which
you never knew where they were going next, and couldn't stop writing
to think about it either. With the short proofs I had time to think
about how I'd attack it as he was writing, and then compare what he'd
written with what I'd written. But I couldn't do that with proofs that
took days and days to elaborate. I couldn't imagine 20 pages of
development down there on the auditorium blackboard in one go, but 5
lines was no trouble.
> In the early days of the Internet where there were only a few
> dozen netnews groups, there were two unix groups only, "unix-wizards"
> and another (the name escapes me). The experts with no patience
> for newbie questions read and posted in the wizards group, those
> who found newbie questions interesting (or amusing) used the
> other. The system seemed to work, maybe its time for
> "comp.os.linux.experts-only"?
It doesn't work for the usual reasons. You know the story. A newsgroup
needs variety in order to survive. Even occasional rows are more
interesting than ask-the-experts round tables. And newbies-only groups
don't work for the obvious reason that it's the blind leading the
blind. A mix is required.
> Just my 2 cents, worth about $0.00000001 on the open market.
Peter
Bill Unruh
>On the other hand, your physix <g> professor must make a choice:
>either teach a simplified physics course that the majority can
>understand and feel good about, or teach a tough course that only
>majors will appreciate. (In my experience about 1 in 30 students.)
>The simple approach is useless to those who plan a career in physics
>and already understand the basics. (Notice how I avoid contention
>by using physics not linux? :-) The expert approach leads to
>student frustation and complaints.
False. I happen to teach physics, including a physics course to
non-science students. You can teach a lot of physics in such a way that
both the mathematically literate and the other students get a lot out of
the course. In fact my arts students solve problems in physics that
stump physics majors. Almost anything can be taught so that both the
experts and the naive learn. It takes more work , and that is why many
teachers take the position that it is impossible. Your so called "expert
approach" usually means the lazy approach.
The same is true for Linux.
>The better schools have two courses, physics for majors and physics
>for non-majors. But if your institution only has one course and
>you must teach it, then you must choose which audience to teach
>for. There is no approach that will satisfy all students.
There is an approach which will teach physics (not mathematics, but
physics) to most of them.
Peter T. Breuer
> Believe me, his bug reports are no more useful. When half a dozen
> different complaints are raised in the one long rant of a report (e.g.
> Debian bug #47000), the inclination is to fix something else instead.
That was uncalled for, colin. Bug 47000 looks kosher to me!
Yes, I don't like man-db. A db for caching manpages isn't neccessary in
these days of fast cpus and fast disks. It's complication that I feel
is unneeded. And there was a time when debian man-db plain didn't work
(it seems better nowadays) in a networked environment. Plus, the
instructions were incomprehensible.
The report in question was headed:
Subject: man-db tries to write in read-only partitions
To: sub...@bugs.debian.org
Date: Sat, 9 Oct 1999 13:12:37 +0200 (MET DST)
And pointed out that the man-db package at that time (1999) tried to
write in the /usr hierarchy from time to time. Boom. That seems to
me to be clear enough as a bug report!
I said that it wasn't a very good advert for debian to have the
man pages not working.
It is very "unsightly" to users. It undoes the work of persuading them
to read manpages in the first place.
I still see problems with man-db. I have no idea why, but occasionally
it just won't see manpages that I can read with nroff. The only cure is
to su - man, and run man-db.
Have a look at the report and see if it makes more sense to you now!
Are you the maintainer? I remember a conversation partly in german
with the author, whoever it is.
I went on to complain about the documentation, pointing out what made
it so wonderfully incomprehensible. Has that been fixed ...
> 'tar cf - ... | ssh remote-host tar xvf -' can be a useful way to get
> around scp's (apparent) lack of an escape.
Yes, good idea. But if you're scp'ying a directory, there won't be any
trouble with the filenames inside.
Peter
Colin Watson
>> Believe me, his bug reports are no more useful. When half a dozen
>> different complaints are raised in the one long rant of a report (e.g.
>> Debian bug #47000), the inclination is to fix something else instead.
>
>That was uncalled for, colin. Bug 47000 looks kosher to me!
It is a kosher bug, yes. Unfortunately, it's also several bugs (some of
which I agree with and some I don't) rolled up into one, together with
lots of sarcastic remarks. Nothing can be closed until the whole thing
is done, which makes tracking very difficult for the maintainer (i.e. I
can't do what I do for every other bug, put closes: #nnnnn against the
fix in the changelog, and have the bug tracking system remember the rest
of it for me).
And, really - if you had thirty-odd open bug reports, and one of them
contained stuff like "What does it mean if not nothing? Please remove
this GUFF!", would you be inclined to fix that one first? It's the best
way of making a non-critical bug fall to the bottom of the pile.
I guess I'm trying to make a point (well, perhaps overstated - I
apologize for going off half-cocked) about bug reporting, just as about
helping people on newsgroups - making sarcastic remarks about the
package or its maintainer is almost always the least productive thing
it's possible to do, particularly when the maintainer is a volunteer.
http://www.chiark.greenend.org.uk/~sgtatham/bugs.html - I get it wrong
from time to time as well, but it's worth remembering that there's an
ordinary person at the other end (usually) trying to fix it.
> It is very "unsightly" to users. It undoes the work of persuading them
> to read manpages in the first place.
>
>I still see problems with man-db. I have no idea why, but occasionally
>it just won't see manpages that I can read with nroff. The only cure is
>to su - man, and run man-db.
'man --debug' is the key to a useful bug report here. Some code related
to this will be changing soon anyway.
man-db is indeed broken in many interesting ways. I hope and believe
that it's a lot less broken now than it used to be (assuming you're
running testing/unstable).
>Have a look at the report and see if it makes more sense to you now!
>Are you the maintainer? I remember a conversation partly in german
>with the author, whoever it is.
The previous maintainer passed away a couple of months ago, and I took
the package over then. Believe me, I have read the report recently. :)
>I went on to complain about the documentation, pointing out what made
>it so wonderfully incomprehensible. Has that been fixed ...
I rewrite bits and pieces as I see them. On one point I agree with
Fabrizio - since I have lots of code to fix and/or rewrite, any offers
to help improve the documentation will be appreciated. Note that
Fabrizio's comment that the manual only has a verbatim licence is
inaccurate, so the whole thing doesn't need to be rewritten.
>> 'tar cf - ... | ssh remote-host tar xvf -' can be a useful way to get
>> around scp's (apparent) lack of an escape.
>
>Yes, good idea. But if you're scp'ying a directory, there won't be any
>trouble with the filenames inside.
Assuming that the directory name doesn't contain colons, of course, and
you could always call tar on a single filename too.
Cheers [followups set to poster, as this is getting onto a different
topic],
--
Colin Watson [cj...@flatline.org.uk]
"The game is on again / A lover or a friend
A big thing or a small / The winner takes it all" - Abba
Peter T. Breuer
> Peter T. Breuer <p...@oboe.it.uc3m.es> wrote:
>>In comp.os.linux.setup Colin Watson <cj...@flatline.org.uk> wrote:
>>> Debian bug #47000), the inclination is to fix something else instead.
>>That was uncalled for, colin. Bug 47000 looks kosher to me!
> It is a kosher bug, yes. Unfortunately, it's also several bugs (some of
> which I agree with and some I don't) rolled up into one, together with
> lots of sarcastic remarks. Nothing can be closed until the whole thing
I don't find the remarks sarcastic, rather more a transcription of
what the students said to me in class!
> is done, which makes tracking very difficult for the maintainer (i.e. I
> can't do what I do for every other bug, put closes: #nnnnn against the
> fix in the changelog, and have the bug tracking system remember the rest
> of it for me).
If you fixed the bug of trying to write in /usr, and made token changes
to the docs to make them more comprehensible, call it closed! You have
my "permission", should you need it! Man-db does work nowadays. It's
stopped behaving strangely (mostly), so there's no need to try and
fathom the docs any more. Doc bug avoided! And if I disagree with the
bug closure I'll just make life hell by entering another bug report,
and we go round again :-). But it's not likely that I'll disagree.
BTw, I see only two bugs reported in that report (my numerals):
(1)
/usr is mounted readonly, yet man insists at times on trying to update
links and caches and dbs. It should check that the dir is writable
before trying it!
(2)
Secondly: many doc bugs. What does "considered \"global\"" in the
manpath.config connote, if anything. Why is it considered? Isn't
it? Is there anything apart from "global", whatever you mean by it?
What does it mean if not nothing? Please remove this GUFF!
> And, really - if you had thirty-odd open bug reports, and one of them
> contained stuff like "What does it mean if not nothing? Please remove
> this GUFF!", would you be inclined to fix that one first? It's the best
Well, I am - if nothing else - qualified to comment on scientific writing!
So perhaps you should take the criticism of the writing seriously.
I took the time to say what was wrong with it! The point (2)
above is the start of the list of specific comments I made about the
writing, quoting paragraphs and commenting on each. In other words, I
treated it as if it were a scientific paper, applying the standard
refereeing practices, the intent of which is to produce a better
document through negotiation with the author ("peer review" :_).
> way of making a non-critical bug fall to the bottom of the pile.
The write on /usr problem was functionally critical. It stopped man-db
working in environments in which /usr is readonly. That the docs were
incomprehensible I'd count as critical in other ways, given that we're
talking about the manual pages package - after all, if they couldn't
explain themselves, they couldn't explain anything.
> I guess I'm trying to make a point (well, perhaps overstated - I
> apologize for going off half-cocked) about bug reporting, just as about
> helping people on newsgroups - making sarcastic remarks about the
> package or its maintainer is almost always the least productive thing
I had no sarcastic intent then. I can't even see any sarcasm in the report
you mentioned now. Are you perhaps reading something into it? I made
the debian report after talking directly with the author (I believe I
recall). Our discussions were fruitful (not fruity!), and I added the
debian report as a sort of deadline/guarantee that something would be
done.
> it's possible to do, particularly when the maintainer is a volunteer.
> http://www.chiark.greenend.org.uk/~sgtatham/bugs.html - I get it wrong
> from time to time as well, but it's worth remembering that there's an
> ordinary person at the other end (usually) trying to fix it.
>> It is very "unsightly" to users. It undoes the work of persuading them
>> to read manpages in the first place.
>>
>>I still see problems with man-db. I have no idea why, but occasionally
>>it just won't see manpages that I can read with nroff. The only cure is
>>to su - man, and run man-db.
> 'man --debug' is the key to a useful bug report here. Some code related
> to this will be changing soon anyway.
> man-db is indeed broken in many interesting ways. I hope and believe
> that it's a lot less broken now than it used to be (assuming you're
> running testing/unstable).
No, I'm running "stable". Perhaps I'll try testing's source.
>>Have a look at the report and see if it makes more sense to you now!
>>Are you the maintainer? I remember a conversation partly in german
>>with the author, whoever it is.
> The previous maintainer passed away a couple of months ago, and I took
> the package over then. Believe me, I have read the report recently. :)
>>I went on to complain about the documentation, pointing out what made
>>it so wonderfully incomprehensible. Has that been fixed ...
> I rewrite bits and pieces as I see them. On one point I agree with
> Fabrizio - since I have lots of code to fix and/or rewrite, any offers
> to help improve the documentation will be appreciated. Note that
I believe I did that, in a very concrete manner.
> Fabrizio's comment that the manual only has a verbatim licence is
> inaccurate, so the whole thing doesn't need to be rewritten.
>>> 'tar cf - ... | ssh remote-host tar xvf -' can be a useful way to get
>>> around scp's (apparent) lack of an escape.
>>
>>Yes, good idea. But if you're scp'ying a directory, there won't be any
>>trouble with the filenames inside.
> Assuming that the directory name doesn't contain colons, of course, and
> you could always call tar on a single filename too.
> Cheers [followups set to poster, as this is getting onto a different
> topic],
I unset them, probably. Dunno why.
Thanks for the followup (honestly!).
Peter
Colin Watson
>In comp.os.linux.setup Colin Watson <cj...@flatline.org.uk> wrote:
>> is done, which makes tracking very difficult for the maintainer (i.e. I
>> can't do what I do for every other bug, put closes: #nnnnn against the
>> fix in the changelog, and have the bug tracking system remember the rest
>> of it for me).
>
>If you fixed the bug of trying to write in /usr, and made token changes
>to the docs to make them more comprehensible, call it closed! You have
>my "permission", should you need it! Man-db does work nowadays. It's
>stopped behaving strangely (mostly), so there's no need to try and
>fathom the docs any more. Doc bug avoided! And if I disagree with the
>bug closure I'll just make life hell by entering another bug report,
>and we go round again :-). But it's not likely that I'll disagree.
Well, it turns out I've done this now (I decided that if I was going to
be ranting on Usenet I might as well take some time off to write the
code ...), so I'll close it in my next upload.
>BTw, I see only two bugs reported in that report (my numerals):
>
>(1)
> /usr is mounted readonly, yet man insists at times on trying to update
> links and caches and dbs. It should check that the dir is writable
> before trying it!
>
>(2)
> Secondly: many doc bugs. What does "considered \"global\"" in the
> manpath.config connote, if anything. Why is it considered? Isn't
> it? Is there anything apart from "global", whatever you mean by it?
> What does it mean if not nothing? Please remove this GUFF!
I see others:
(3) No way to tell man to just get on with it, and not worry about
bringing the database up to date;
(4) Manpaths in the configuration file aren't reordered, so you have to
put deeper manpaths in the filesystem hierarchy before shallower
ones;
(5) (implied by the discussion) The whole path -> manpath mapping is
just hideously confusing at times.
Feel free to file separate bugs about those. The last will probably just
have to be fixed by better documentation.
>> And, really - if you had thirty-odd open bug reports, and one of them
>> contained stuff like "What does it mean if not nothing? Please remove
>> this GUFF!", would you be inclined to fix that one first? It's the best
>
>Well, I am - if nothing else - qualified to comment on scientific writing!
>So perhaps you should take the criticism of the writing seriously.
>I took the time to say what was wrong with it! The point (2)
>above is the start of the list of specific comments I made about the
>writing, quoting paragraphs and commenting on each. In other words, I
>treated it as if it were a scientific paper, applying the standard
>refereeing practices, the intent of which is to produce a better
>document through negotiation with the author ("peer review" :_).
Oh, I appreciate criticism and take it seriously - note, though, that
neither Fabrizio nor I wrote that in the first place (the text was
direct from 2.3.10, circa 1995), so the more vitriolic bits of
second-person criticism are misdirected. I suppose I should just develop
a thicker skin and get used to it, as people often conflate the author
and the maintainer ... Anyway, I've rewritten the paragraphs in
question, and they should make more sense now.
>> way of making a non-critical bug fall to the bottom of the pile.
>
>The write on /usr problem was functionally critical. It stopped man-db
>working in environments in which /usr is readonly.
I was thinking versus security problems and the like, of which man-db
has had more than its fair share recently. However, when I fixed man to
cope gracefully with permission problems in certain cases, I discovered
that it was key to lots of other stuff, so I owe you an apology. You
should get a copy of the changelog in a couple of days when I close that
bug and quite a few others.
>> I guess I'm trying to make a point (well, perhaps overstated - I
>> apologize for going off half-cocked) about bug reporting, just as about
>> helping people on newsgroups - making sarcastic remarks about the
>> package or its maintainer is almost always the least productive thing
>
>I had no sarcastic intent then. I can't even see any sarcasm in the report
>you mentioned now. Are you perhaps reading something into it? I made
>the debian report after talking directly with the author (I believe I
>recall). Our discussions were fruitful (not fruity!), and I added the
>debian report as a sort of deadline/guarantee that something would be
>done.
Perhaps so - normally I can read past differences in writing style, but
not always. Since the previous maintainer said at one point "just
guessing, no need to be flamed again", I took it that he was feeling
under some pressure as well.
>>>I went on to complain about the documentation, pointing out what made
>>>it so wonderfully incomprehensible. Has that been fixed ...
>
>> I rewrite bits and pieces as I see them. On one point I agree with
>> Fabrizio - since I have lots of code to fix and/or rewrite, any offers
>> to help improve the documentation will be appreciated. Note that
>
>I believe I did that, in a very concrete manner.
Your side of the conversation hasn't been recorded in the bug log,
except by way of quotes in replies, so that wasn't clear to me. The
current source is in CVS at http://sourceforge.net/projects/man-db/; I
can give you write access.
>> Cheers [followups set to poster, as this is getting onto a different
>> topic],
>
>I unset them, probably. Dunno why.
Well, followups away from .networking, anyway ...
>Thanks for the followup (honestly!).
It appears to have borne some useful fruit, to my own surprise as much
as anyone else's. :)
Cheers,
--
Colin Watson [cj...@flatline.org.uk]
"Everyone, please welcome our new friend Stef. He's here with us
because he thinks he's a penguin." - http://www.userfriendly.org/