davidpelfree / pipeviewer Goto Github PK

What steps will reproduce the problem?
1. ./configure
2. make
3. (complain loudly)

What is the expected output? What do you see instead?

What version of the product are you using? On what operating system?
Trunk. OS X 10.4.

Please provide any additional information below.

This fixed my build problem. I came to pv from MacPorts, but wanted to fix it 
with a patch 
upstream.

See what happens if we fsync() and/or use O_SYNC, which would be useful when 
writing to USB devices or other things that tend to hang for ages after we've 
finished. (Original request submitted by an anonymous user)

It has been suggested that using splice() on Linux would boost performance 
(Коренберг Марк / socketpair).

See email reports from Bob Friesenhahn.

Patch from Sami Liedes.

Provide an "-F" option to allow pv to be given a list of files to read from
the given file. Provide an additional option for that list to be
null-terminated instead of newline-terminated.

Statistics like average(mean), min/max throughput rate and standard
deviation would be useful to have.

I think this would be a very useful feature allowing to validate the data 
integrity on the receiving end if e.g. data is just dumped into a file.  E.g. 
if I want to backup a partition of a drive, as is, running twice checksum first 
and then piping through pv would be evil, and constructing evolved shell 
commands like

$> cat ~/.emacs > >(md5sum - >| /tmp/1) | pv > /dev/null 
71,2kB 0:00:00 [  13MB/s] [  $> cat /tmp/1
e81dbdf1a7a273a60b789c55e8ebf65e  -
$> md5sum .emacs
e81dbdf1a7a273a60b789c55e8ebf65e  .emacs

would be ugly and error prone

What is the expected output?

upon exit print checksums of the "streamed" data, e.g.

MD5: e81dbdf1a7a273a60b789c55e8ebf65e
SHA: 0960cc9bdebeebd590c39cfc9b29f36530dc223e

What do you see instead?

nothing :-/

What version of the product are you using? On what operating system?

1.2.0-1 debian sid amd64

Please provide any additional information below.

I love beer

And thanks for the pv -- it is really nice ;)

Add -O to allow ordering of -pterb output, eg -bertp (Vladimir Pal) 
(http://unixstuff.ru), idea by Vladimir Ermakov

If the rate is limited to less than 10 bytes per second with the default 
interval, no data moves anywhere. See email report from Henry Precheur.

What steps will reproduce the problem?
1. try to cross-compile for uClinux. cross-compilation for other platforms may 
fail as well - just can't test it right now.
2. compilation fails due to the host ld being invoked (instead of 
bfin-uclinux-ld).
3. scratch your head while reading autoconf/make/vars.mk, line 33.

What is the expected output? What do you see instead?
Expected to see a cross-compiled version of pv, instead got a linking error due 
to host ld trying to process cross-compiled object files.

What version of the product are you using? On what operating system?
Using version 1.2.0 on Linux Mint Debian Edition, though it really does not 
matter in this case.

Please provide any additional information below.
The fix is very simple - just remove that line (with "LD = ld") from vars.mk, 
re-run generate.sh, configure, make. Now the correct version of ld is invoked. 
Got this solution from here: http://codereview.chromium.org/3907003

It's useful to use the terminal title line to report progress, so that "normal" 
stderr (and stdout) 
reporting can also be used. A simple patch is provided, which when given the 
"-x" option writes 
the required ansi escape codes around the progress output.

because  de for  sentence  is missing $@


for POFILE ; do
...
done

must be 
for POFILE in $@; do
...
done

to replicate delete  src/nls/table.c and run  make 

Integrate attached patch from E. Rosten.

On OSes without compressed man pages, installation fails. See email report from 
Bob Friesenhahn.

What steps will reproduce the problem?
1. Build pv 1.1.4 under OS X 10.4.11 (the patch from issue 5 is needed)
2. make test

What do you see instead?
010-dd: FAILED
make: *** [test] Error 1

If I uncomment the echo from the test, and remove the -q flag from pv, it 
appears to not be providing enough data to dd:
010-dd: 62.5MB 0:00:00 [99.4MB/s] 
[<=>                                                ]
[100000000] [ 65503264]
FAILED
make: *** [test] Error 1

What version of the product are you using? On what operating system?
pv-1.1.4, OS X 10.4.11, GCC i686-apple-darwin8-gcc-4.0.1

What steps will reproduce the problem?
1. Lookup the PID of some process running on your system.
2. Run: "pv -R PID"

What is the expected output? What do you see instead?

I expect pv to print out an error that the designated PID is not an instance of 
pv and then exit without sending a signal to that PID. Instead pv will send the 
SIGUSR1 signal to the process regardless of what program it actually is.

What version of the product are you using? On what operating system?

1.1.4-1 on debian

Please provide any additional information below.

You could either check the output "basename /proc/PID/exe", or "ps -o comm= 
PID". And verify that it is "pv" before sending the signal. I don't know which 
approach is more portable.

What steps will reproduce the problem?
1. Open a terminal window with rxvt on Cygwin
2. tar c subdir/ | pv -pterW -s 99999999 -N 'Source' | bzip2 | pv -pterW -N
'Compressed' > tempo.tbz2

What is the expected output?
Like the demo pictures.

What do you see instead?
The lines overwrite each other as if they both do CR but don't move back
up. This does not only happen if it is at the bottom of the window. Adding
"-c" option results in "failed to lock terminal: Invalid argument"

What version of the product are you using? 1.1.0

On what operating system? 
$ uname -a
CYGWIN_NT-5.1 brisingamen 1.5.24(0.156/4/2) 2007-01-31 10:57 i686 Cygwin

Currently, the cursor positioning code uses locking on the terminal, or a
lockfile when O_EXLOCK is unavailable, to manage concurrent access so that
the different instances of pv can all output to the same terminal.

Instead of using this method a semaphore or other shared memory type
solution should be used, since shared memory is already being used for
other aspects of the cursor positioning code anyway and it would be cleaner
than relying on file locks.

Just downloaded the cygwin and selected current (stable) version of packages, 
but no matter how much I've searched, couldn't find on it the pv command.

I am checking block device backups with cmp, and using pv to watch progress.

cmp has a skip option that allows to checking a problem area again, but is hard 
to combine with pv. For example, doing

    dd ibs=1c skip=$SKIP if=$FILE1 |pv |cmp -i $SKIP $FILE2

works, but only with 1-byte blocks, and very slowly. With a `--skip` option, 
things would work much better:

    pv --skip=$SKIP $FILE1 |cmp -i $SKIP $FILE2

pv can skip efficiently with a seek, and know how much data is still left to 
compute an ETA.

cmp also supports a --bytes option:

    dd ibs=1c skip=$SKIP count=$BYTES if=$FILE1 |pv |cmp -i $SKIP --bytes $BYTES $FILE2

Which stops comparing after reading so many bytes. It doesn't really need to be 
supported by pv, since cmp will close stdin when it has read enough. Adjusting 
the size estimate with -s is enough:

    pv --skip=$SKIP -s $BYTES $FILE1 |cmp -i $SKIP --bytes $BYTES $FILE2

In a multi-pv pipeline, if the first pv exits, another should take over the
IPC leader role.

Option for process title (Martin Sarsale)as "pv - name:FooProcess -xyz - 
transferred: 1.3GB - 500KB/s - running: 10:15:30s"