[yt-users] Installing yt without hg

Tue Nov 29 02:09:20 PST 2011

Dear all,

I have the same issue with yt, on a machine with a very strict policy on 
Internet access. I solved it by remotely mounting the yt directory on my PC, 
and running the install_script partly on my PC, and partly on the original 
machine.  In fact, I run all the downloads and mercurial-related parts on my 
PC, and the installations on the remote machine. It is rather lengthy and 
fragmented, but it works. I just comment the pip lines, too. 

The solution proposed by Dave sounds very useful, especially for people 
running the stable version, or less interested to updates in the source.

  Luigi

On Tuesday 29 November 2011, David Collins wrote:
> I have now been successful.  After getting rid of pip, everything
> worked.  I think it's as easy as
> 1.) grab all the tar files on a machine with access (all the
> dependancies in  .tar.bz that install.sh grabs, plus a tar of the yt
> repo)
> 2.) get them to the blind machine
> 3.) comment out the pip install
> 4.) bash install
>
> I haven't run into the issue of updating yt, so i may run into the
> problems Matt mentioned about develop, but I made a slice and it looks
> fine.
>
> I don't think it's worth messing with supporting a whole extra
> tarball, since as Cameron mentioned this is not the most common method
> for people getting it, and most of the machinery to do what needs to
> be done is already in that install script.  I think a few little
> tweaks to the install script can make the existing machinery do double
> duty, if it had two extra conditionals:  one "download only" that just
> pulled the dependancy .tar.gz and cloned the repo, and one "blind
> make" that basically did exactly what the script does, but skipped
> pip.  I imagine that pip is of limited utility on such a machine
> anyhow...
>
> I'll report back as I break things.
>
> d.
>
> On Mon, Nov 28, 2011 at 8:00 PM, Matthew Turk <matthewturk at gmail.com> wrote:
> > Hi Elizabeth, and everyone else,
> >
> > A while ago there was some talk of doing a monolithic install.  This
> > would solve the problem of getting the dependency files (which Dave
> > has pretty easily solved; in fact, that's not usually the hard part.)
> > The second issue is not directly related to this, which is that
> > distribute and pip are based on the idea of having network access.
> >
> > Right now the line "easy_install-2.7 pip" is the one causing problems,
> > which can easily be remedied by adding a pip tarball to the install
> > dependency list.  The second issue is that occasionally distribute
> > will try to hit the network.  This will require changes to the file
> > setup.py in yt, which I have not yet experimented which but which I
> > outlined in my email to Dave.
> >
> > Moving forward, a monolithic tarball of dependencies is something I
> > would strongly support; this would need to be managed by someone other
> > than me, cryptographically signed, and would contain everything.  It
> > would still require the modifications I outlined above to the setup.py
> > script.
> >
> > If someone would like to take up this project, it would definitely be
> > welcomed!  I could provide input as could any other interested
> > parties.
> >
> > -Matt
> >
> > On Mon, Nov 28, 2011 at 9:26 PM, Elizabeth Tasker
> >
> > <tasker at astro1.sci.hokudai.ac.jp> wrote:
> >> At least 3 people that I know of have run into this problem so it
> >> seems like it might be the sort of issue that keeps re-occurring? We
> >> have worked around it here now by asking the system administrators of
> >> the machine to install yt (I'm unsure currently how good they'll be at
> >> updates if we ask, but at present it's fine).
> >>
> >> Copying across the tarball does work, but you still need to tweak the
> >> install script to get it to compile. I did this OK, but I remember it
> >> not being a super quick job. Possibly an alternative install script or
> >> directions for copying and modification would work just as well?
> >>
> >> Elizabeth
> >>
> >> On 29 November 2011 11:12, Cameron Hummels <chummels at astro.columbia.edu> 
wrote:
> >>> Are there that many people who need this sort of thing?  It was my
> >>> understanding that most people could use mercurial to get the code
> >>> without problems.  In the worst case, you could download the code on a
> >>> machine that *did* have access, and then tar it up yourself and send it
> >>> to the machine that didn't have access.
> >>>
> >>> Perhaps it isn't a bad idea to have a tarball for each point release?
> >>>  I just don't want to make a bunch of extra work if it isn't actually
> >>> going to get used.
> >>>
> >>> Cameron
> >>>
> >>> On 11/28/11 8:58 PM, Elizabeth Tasker wrote:
> >>>> I also had to install yt on a machine done up like Alcatraz. I don't
> >>>> think I had the same error as Dave (it was a while ago) but it did
> >>>> take me a while and a great deal of frustration.
> >>>>
> >>>> Could we maybe consider a tarball for yt that doesn't require any
> >>>> internet access? Even if it's not updated all that often?
> >>>>
> >>>> Elizabeth
> >>>>
> >>>> On 29 November 2011 09:29, David Collins<dcollins at physics.ucsd.edu>
> >>>>
> >>>>  wrote:
> >>>>> This particular failure is on the installation of pip.  I commented
> >>>>> that out and it looks like numpy has gone just fine and mpl is going
> >>>>> fine, but I imagine when I get to the setuptools that will cause
> >>>>> problems, as well.  More then it gets there.
> >>>>>
> >>>>> d.
> >>>>>
> >>>>> On Mon, Nov 28, 2011 at 5:00 PM, Matthew Turk<matthewturk at gmail.com>
> >>>>>
> >>>>>  wrote:
> >>>>>> Hi Dave,
> >>>>>>
> >>>>>> Is it dying inside setup.py in yt?  If so, try disabling that in
> >>>>>> setup.py (lines 5&6).  It might break, in which case, we can
> >>>>>> probably get rid of the distribute/setuptools specific options.  I
> >>>>>> think this would be the entry_points option (lines 122-124) and
> >>>>>> possibly line 130.  These options provides the 'yt' command, which
> >>>>>> maybe we can mock up using a script rather than an entry point, and
> >>>>>> then get rid of setuptools altogether.  You will also not be able to
> >>>>>> use the "develop" command if you do not use setuptools/distribute,
> >>>>>> in which case you'll be stuck installing and having to reinstall if
> >>>>>> you make changes.
> >>>>>>
> >>>>>> I strongly encourage you to file as a bug with the distribute folks
> >>>>>> that you can't install a distribute-bootstrapped package without a
> >>>>>> network connection.
> >>>>>>
> >>>>>> -Matt
> >>>>>>
> >>>>>> On Mon, Nov 28, 2011 at 6:49 PM, david collins<antpuncher at gmail.com>
> >>>>>>
> >>>>>>  wrote:
> >>>>>>> Hi there--
> >>>>>>>
> >>>>>>> I'm on a system that has a pretty fierce firewall so that things
> >>>>>>> like pip, wget, and mercurial can't see the outside world through
> >>>>>>> any http based protocol.  I got the zips from a different machine
> >>>>>>> and things worked basically fine until I got to the bit where it
> >>>>>>> tries to install easy_install.  (error below)
> >>>>>>>
> >>>>>>> Has anyone had experience with such a problem, and have suggestions
> >>>>>>> for getting around the need for a network connection?
> >>>>>>>
> >>>>>>> Thanks,
> >>>>>>> d.
> >>>>>>>
> >>>>>>> Installing easy_install script to
> >>>>>>> /usr/projects/magnetic/david_c/local/bin
> >>>>>>> Installing easy_install-2.7 script to
> >>>>>>> /usr/projects/magnetic/david_c/local/bin
> >>>>>>>
> >>>>>>> Installed
> >>>>>>> /usr/projects/magnetic/david_c/local/lib/python2.7/site-packages/di
> >>>>>>>stribute-0.6.21-py2.7.egg Processing dependencies for
> >>>>>>> distribute==0.6.21
> >>>>>>> Finished processing dependencies for distribute==0.6.21
> >>>>>>> After install bootstrap.
> >>>>>>> Creating
> >>>>>>> /usr/projects/magnetic/david_c/local/lib/python2.7/site-packages/se
> >>>>>>>tuptools-0.6c11-py2.7.egg-info Creating
> >>>>>>> /usr/projects/magnetic/david_c/local/lib/python2.7/site-packages/se
> >>>>>>>tuptools.pth Searching for pip
> >>>>>>> Reading http://pypi.python.org/simple/pip/
> >>>>>>> Download error on http://pypi.python.org/simple/pip/: timed out --
> >>>>>>> Some packages may not be found!
> >>>>>>> Reading http://pypi.python.org/simple/pip/
> >>>>>>> Download error on http://pypi.python.org/simple/pip/: timed out --
> >>>>>>> Some packages may not be found!
> >>>>>>> Couldn't find index page for 'pip' (maybe misspelled?)
> >>>>>>> Scanning index of all packages (this may take a while)
> >>>>>>> Reading http://pypi.python.org/simple/
> >>>>>>> Download error on http://pypi.python.org/simple/: timed out -- Some
> >>>>>>> packages may not be found!
> >>>>>>> No local packages or download links found for pip
> >>>>>>> error: Could not find suitable distribution for
> >>>>>>> Requirement.parse('pip')
> >>>>>>>
> >>>>>>>
> >>>>>>> --
> >>>>>>> Sent from my computer.
> >>>>>>> _______________________________________________
> >>>>>>> yt-users mailing list
> >>>>>>> yt-users at lists.spacepope.org
> >>>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> yt-users mailing list
> >>>>>> yt-users at lists.spacepope.org
> >>>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>>>>
> >>>>> --
> >>>>> Sent from my computer.
> >>>>> _______________________________________________
> >>>>> yt-users mailing list
> >>>>> yt-users at lists.spacepope.org
> >>>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>>>
> >>>> _______________________________________________
> >>>> yt-users mailing list
> >>>> yt-users at lists.spacepope.org
> >>>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>>
> >>> _______________________________________________
> >>> yt-users mailing list
> >>> yt-users at lists.spacepope.org
> >>> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >>
> >> _______________________________________________
> >> yt-users mailing list
> >> yt-users at lists.spacepope.org
> >> http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org
> >
> > _______________________________________________
> > yt-users mailing list
> > yt-users at lists.spacepope.org
> > http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

-- 

---------------------------------------------------------------

Luigi Iapichino
Zentrum für Astronomie der Universität Heidelberg
Institut für Theoretische Astrophysik
Albert-Ueberle-Str. 2, D-69120 Heidelberg, Germany
Tel: +49 6221 548983, Fax: +49 6221 544221
e-mail: luigi at uni-heidelberg.de
URL: http://www.ita.uni-heidelberg.de/~luigi/