Python extra verbose in 80x IBs

[davidlange6]

[dlange@cmsdev01 src]$ python
Python 2.7.11 (default, Jan 21 2016, 17:19:47)
[GCC 5.3.0] on linux2
Type "help", "copyright", "credits" or "license" for more information.
Name collision 'vt52' between
vt52|dec vt52
and vt52|dec vt52
...now dec vt52
Cannot remove alias 'dec vt52'
Name collision 'vt100-nav' between
vt100-nav|vt100 without advanced video option
and vt100-nav|vt100 without advanced video option
...now vt100 without advanced video option
Cannot remove alias 'vt100 without advanced video option'
Name collision 'vt102' between
vt102|dec vt102
and vt102|dec vt102
...now dec vt102
Cannot remove alias 'dec vt102'
Name collision 'vt100' between
vt100|vt100-am|dec vt100 (w/advanced video)
and vt100|vt100-am|dec vt100 (w/advanced video)
...now vt100-am|dec vt100 (w/advanced video)
...now dec vt100 (w/advanced video)
Cannot remove alias 'dec vt100 (w/advanced video)'
Name collision 'vt220' between
vt220|vt200|dec vt220
and vt220|vt200|dec vt220
...now vt200|dec vt220
...now dec vt220
Cannot remove alias 'dec vt220'
Name collision 'ansi' between
ansi|ansi/pc-term compatible with color
and ansi|ansi/pc-term compatible with color
...now ansi/pc-term compatible with color
Cannot remove alias 'ansi/pc-term compatible with color'
Name collision 'linux' between
linux|linux console
and linux|linux console
...now linux console
Cannot remove alias 'linux console'
Name collision 'dumb' between
dumb|80-column dumb tty
and dumb|80-column dumb tty
...now 80-column dumb tty
Cannot remove alias '80-column dumb tty'

[davidlange6]

assign core

[cmsbuild]

New categories assigned: core

@Dr15Jones,@smuzaffar you have been requested to review this Pull request/Issue and eventually sign? Thanks

[cmsbuild]

A new Issue was created by @davidlange6 (David Lange).

@davidlange6, @smuzaffar, @Degano, @davidlt, @Dr15Jones can you please review it and eventually sign/assign? Thanks.

cms-bot commands are listed here

[bendavid]

I have also seen this running madgraph btw (which runs python underneath anyway), though it seems to have no ill effect beyond the extra output.

[davidlt]

Went to cmsdev01 and executed python, didn't get this output. Looks like some internal failures/assert/unknown state maybe related to term cap database.

[cbernet]

Hi David, all, I got here by googling this printout :-) I also have the problem.

[davidlt]

It's due to terminfo/termcap not matching up on your system and what's in CMSSW.

[davidlt]

I might fix it with arrival of GCC 6.1.0, but I am still debating with myself.

[bendavid]

Well this occurs on lxplus, which I think can safely be considered a "reference" system for running CMSSW.

[davidlt]

The build is never API/ABI stable and is never fully relocatable. We never test if packages we build pass tests (e.g. we don't even know if compiler, i.e. GCC, generates code in expected way). Doing dnf/yum upgrade/update can break it. Which recently happened on Fedora 22 builds where NSS was updated and finally it wasn't compatible with what we shipped as part of bootstrap-bundle package. You couldn't even bootstrap anymore because RPM wasn't working due to broken NSS setup.

Things are fragile, quite stable of ancient beasts as RHEL/CentOS as there are strict rules on what can/cannot be updated/modified meaning that API/ABI usually will be stable, but not behaviour (.e.g, sometimes getting 5 times slower std::log is expected after upgrade/update).

In this case it's relatively easy to fix. I just remove ncurses package from our distribution and that is fine because we always requested from ncurses to be installed on the system. So, we install ncurses-devel instead on our build machines. But in order to do that, I have rebuild the bootstrap itself. Which is easiest with GCC 6.1.0, because it's soon to arrive. But yes, it can also be done for GCC 5.3.0.

I used to play with ncurses termcap and terminfo, but I never managed to get all externals happy with modifications. Removing ncurses from distribution is the best for many reasons.

[rijuvenator]

Hi,

Was this problem ever solved? It still seems to be happening on lxplus, using CMSSW_8_1_0_pre8

[bartosik-hep]

@davidlt, @davidlange6 I have the same issue. There is no 8X release at the moment running on GCC greater than 5.3.0. Is there any way to suppress the message or to solve the issue in a 8X release?

[rijuvenator]

If you don't care about errors, simply do

python script.py 2>/dev/null

If you care about errors, I have a workaround.

Put this in your bash profile:

exec 3>&1

Then, anytime you run a Python script with, for example,

import ROOT as R

replace it with this:

import sys, os
import ROOT as R
fd3 = os.fdopen(3, 'w')
sys.stderr = fd3

Then you can run all your Python scripts with

python script.py 2>/dev/null

This will redirect the ROOT error messages to /dev/null, then set stderr to point to file descriptor 3, so that you still see all Python error messages (because they will get redirected to the terminal), but you won't see the ROOT error messages.

[MengqingWu]

@davidlange6 I have the same issue with 8_0_11, but can not remember if it existed from the beginning or not. It shows today..

[binka04]

Hii, davidlange6,
Has the problem been fixed ? , I am using CMSSW_8_1_0_pre9, slc6_amd64_gcc530. I am seeing such kind of error message while running a python script in cmslpc account.
The python script import the following.
from os import environ
from ROOT import *
from math import sqrt
gROOT.ProcessLine(".L ~/tdrstyle.C")
gROOT.ProcessLine("setTDRStyle()")
from optparse import OptionParser

The link to python script is here:
https://github.com/awhitbeck/RA2bZinvEst/blob/RA2b_V12_v0.1.2/src/formatIntegrationInputs.py

and I am attaching output message in a pdf file here. It will be great help to me, if you can suggest me what is the reason of bug.

[Thanks.]
ouput.pdf