I'd try the following:

* get a copy of the website locally

* make all the symlinks relative (is it OK?)

* check in all the non-binaries and symlinks

Are there any symlinks to binaries?


On Wed, Sep 7, 2016 at 3:32 PM, Harald Schilly <[email protected]> wrote:
> Hi, I started https://github.com/sagemath/files and already added a
> bit on the server, but stumbled over general idiocies. So, I don't
> want to spend more time on this. E.g. there are symlinks, but git
> doesn't like them, and there is something odd going on with an already
> existing ssh key vs. github. Maybe with a little bit more nerves I am
> able to do this. (unless I really need to bind-mount directories as
> root, that might then be impossible for me)
>
> -- h
>
>
> On Sun, Sep 4, 2016 at 12:28 PM, Dima Pasechnik <[email protected]> wrote:
>> Well, I already complained to Harald about another README there that  is
>> hopelessly old too.
>> I'd like to propose putting all these files there on
>> github.com/sagemath/mirrorfiles or something like that
>> and maintain them there.
>>
>>
>>
>> On Saturday, September 3, 2016 at 12:55:36 PM UTC, leif wrote:
>>>
>>> leif wrote:
>>> > John Cremona wrote:
>>> >> On 2 September 2016 at 16:36, leif <[email protected]
>>> >> <mailto:[email protected]>> wrote:
>>> >>
>>> >>     John Cremona wrote:
>>> >>     > Since I had never tried downloading and running a binary, I
>>> >> thought I
>>> >>     > would.  For a laptop running ubuntu 14.04 I looked at the UK
>>> >> mirror and
>>> >>     > found no 7.3 binary so I downloaded the 7.2 one (there was 7.3
>>> >> for
>>> >>     > ubuntu 12.04 but not 14.04 or later).
>>> >>
>>> >>     32-bit?!?!!!!  (For Sage 7.3, there are 64-bit binaries for 12.04,
>>> >>     14.04, 15.10 and 16.04.)
>>> >>
>>> >>
>>> >> Well yes (uname -m returns i686).  For some reason I did this
>>> >> experiment
>>> >> on a small and slow Toshiba netbook.
>>> >>
>>> >>
>>> >>
>>> >>     As reported on sage-release, 32-bit (native) Ubuntu builds
>>> >> currently
>>> >>     don't work for any Ubuntu version > 12.04 because of issues with
>>> >>     -fstack-protector (which Ubuntu's GCCs by default enable).  Nobody
>>> >> has
>>> >>     yet tracked this further down.  (I planned to revive a 32-bit
>>> >> machine
>>> >>     for debugging/testing, but haven't yet had the time, but there
>>> >> doesn't
>>> >>     seem to be much demand either.)
>>> >>
>>> >>
>>> >> I had not realised this was such a can of worms.  I used to regularly
>>> >> build Sage on this machine (slowly, but then I do sleep) and last did
>>> >> so
>>> >> with 7.0.   I can do so again if there is call for it (and this
>>> >> conversation is better suited to sage-devel).
>>> >
>>> > Well, give for example Sage 7.3 a try.  In case that works for you
>>> > (without setting SAGE_INSTALL_GCC=yes), you can create a bdist yourself
>>> > (see link below).
>>> >
>>> > I guess my Pentium4 (though with just 2GB, and USB-2.0-attached external
>>> > disk only) would be a bit faster, but I'd have to repair the SFF power
>>> > supply, or rather replace its fan once again; it also at the moment has
>>> > Lucid and Precise only...)
>>> >
>>> >
>>> >>     > Using the command-line I unpacked
>>> >>     > the tarball (tar jxf ...tar.bz2) which created a SageMath
>>> >> directory, so
>>> >>     > I cd'd into there and typed ./sage.  As the original poster
>>> >> reported,
>>> >>     > this resulted in a lot of "patching..." messages appearing,
>>> >> followed by
>>> >>     > the 7.2 banner and a sage: prompt.  Subsequent runs also worked
>>> >> without
>>> >>     > the patching stuff.
>>> >>     >
>>> >>     > This does not help much, though I wonder how many of the posted
>>> >> binaries
>>> >>     > are tested?  And why is it neccessary to patch all those files?
>>> >>
>>> >>     Because unfortunately people decided to break "relocating" Sage,
>>> >> which
>>> >>     still worked a while ago (modulo very few and minor issues
>>> >> perhaps).
>>> >>
>>> >>     So bdists are now made with some separate script / program from
>>> >> Volker,
>>> >>     such that they "patch" themselves upon installation / first attempt
>>> >> to
>>> >>     run 'sage'.  Loads of (absolute) paths in scripts but also binaries
>>> >> and
>>> >>     libraries thereby get (again) hardcoded to the actual installation
>>> >>     folder.
>>> >>
>>> >> I thought that would be the reason;  so it's Volker's script which
>>> >> could
>>> >> be made less frightening to the novice user.
>>> >
>>> > https://github.com/sagemath/binary-pkg
>>> >
>>> > You can create an issue or a pull request... ;-)
>>>
>>> P.S.:  There's a horribly outdated README.txt (with still uppercase
>>> SAGE, and among other flaws, telling one could move the Sage tree
>>> anywhere):  http://files.sagemath.org/linux/32bit/README.txt
>>>
>>> I guess the ones in the other bdist subfolders aren't much better.
>>>
>>> On the other hand, the web page itself (i.e., index.html) gives recent
>>> info on uncompressing the various(?) formats offered.  There we could
>>> also add some short info on how to proceed after downloading.
>>>
>>>
>>> More worms escaping the can...  (I'd say Pandora's box though.)
>>>
>>>
>>> -leif
>>>
>>>
>>> P.P.S.:  Just noticed in the mentioned description of compression
>>> formats, there's "everything" but what we currently solely offer (namely
>>> .tar.bz2, for whatever reason)... 8-)
>>>
>>> Nearly the same for the linux/64bit/ folder, including README.txt (while
>>> there's also some left-over beta rpm, and also an obsolete
>>> sage-x.y.z-sage.math.washington.edu-x86_64-Linux.txt).
>>>
>>> CC'ing Harald, as I'm not going to open an issue on GitHub.  Maybe we
>>> should also forward this to sage-devel, but with a more appropriate title.
>>>
>>>
>>

-- 
You received this message because you are subscribed to the Google Groups 
"sage-support" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/sage-support.
For more options, visit https://groups.google.com/d/optout.

Reply via email to