kcrisman wrote:
On Feb 15, 12:23 pm, "Dr. David Kirkby" <david.kir...@onetel.net>
wrote:
kcrisman wrote:
R's spkg-install needs an overhaul, as noted in my previous email (which It
seems to have invalid options, invalid comments, and various other issues. It
would appear we could potentially get better performance too, at the cost of
installing other libraries.
If you could point those out explicitly on the new ticket you created,
that would be awesome.
Actually, I think I should create a new ticket, for
"R's spkg-install needs a good overhaul"
or something like that, as the ticket I created is very specific to the iconv
code. The problems with R's spkg-install appear to go far beyond that.
With regard to the R graphics ticket, does Solaris automatically (or
Linux, for that matter) have Xwindows support?
Solaris supports X. There is a directory /usr/include/X11, which has loads of X
related files. However, R's spkg-install has these 7 lines of code:
------------------------------------------------------------------
# I have problems with this on OSX Intel 10.5.1 -- for now just turn it off.
# It will be good to get something fully working before worrying about X.
if [ -f /usr/include/X11/Xwindows.h ]; then
XSUPPORT=yes
else
XSUPPORT=no
fi
---------------------------------------------------------------------
I don't have the file /usr/include/X11/Xwindows.h on my Solaris boxes, so X is
disabled. I noticed sage.math does have that file, so I assume X is enabled on
Linux (at least on sage.math).
The fact the problem mentioned in spkg-install appeared to have been on OSX
Intel 10.5.1, has not stopped someone disabling X on every platform if
/usr/include/X11/Xwindows.h does not exist.
I don't know if the file /usr/include/X11/Xwindows.h is mandatory for R to
support X.
See the ticket about R graphics (don't have the # in front of me) for
an update on this, and a link to an spkg. However, I know nothing
about X; the current spkg changes things to use quartz for Mac and
continue to look for this file for Xwindows. Any comments you have
about graphics-specific things like this you should put on that
ticket.
I don't know much about this.
It strikes me that packages get updated, and nobody actually looks *carefully*
at what implications that update might have. In this case, the update broke the
build on Solaris, has useless options, patches that might not be needed, and
might be undesirable .... etc.
Yes, please let us know about the other stuff (problems not related to
iconv or X) on a new ticket.
I created a new ticket, which lists some of the issues I see with R. It needs
someone with more knowledge of R, and who is willing to read the installation
guide, to tackle this.
http://trac.sagemath.org/sage_trac/ticket/8274
<unrelated-to-R>
But in response to the other comments, the reality is that often
someone updating an spkg needs the new functionality on a few
platforms, and knows very little about configuration and/or other
platforms. If it doesn't break something on currently supported
platforms, we seem to go with upgrades.
IMHO, if someone can't get these tested on supported platforms (which includes
Solaris), then it should go into experimental until properly tested.
http://www.sagemath.org/doc/developer/inclusion.html?highlight=solaris
says
"Some Sage developers are willing to help you port to OSX, Solaris and Windows.
But this is no guarantee and you or your project are expected to do the heavy
lifting and also support those ports upstream if there is no Sage developer who
is willing to share the burden."
But it seems to me, that this is not happening. Packages get updated, without
being tested on other platforms, and break builds.
Would it not seem more reasonable that updates to standard packages go into
"optional" until fully tested? Debian do this - they have 3 in fact. Stable,
testing and unstable.
If person X wants some extra functionality, they should be happy to use an
optional package unless they are willing to either test it themselves more
fully, or get acknowledgment from others that it has been tested.
Thanks for your valuable help on this stuff! It will make Sage so
much more stable long-term, I think.
- kcrisman
Well, if Sage is ever to become a viable alternative to the commercial products,
I think it needs more testing. I do not believe Wolfram Research would compile
Mathematica on all platforms they support (which includes Solaris), and never
test it on Solaris before making a new release!
Dave
--
To post to this group, send an email to sage-devel@googlegroups.com
To unsubscribe from this group, send an email to
sage-devel+unsubscr...@googlegroups.com
For more options, visit this group at http://groups.google.com/group/sage-devel
URL: http://www.sagemath.org
