From: Conrad S. <co...@he...> - 2000-12-01 19:14:12
|
Hi This is the first post of the SciGraphica compile FAQ that will hopefully answer some questions. It is in the current CVS. --------------------------------------------------------------------------- If you have trouble compiling SciGraphica, please read these FAQs and answers first. Version 0.01 Conrad Steenberg 1/12/2000 1.0 I can't compile SciGraphica, it seems really hard. A: Look around for binary packages for your distribution. We have posted binaries on http://sourceforge.net/projects/scigraphica/ for redhat 6.x on Linux x86 and Alpha. 1.1 What packages do I need for _optimal_ usage/compilation of Scigraphica? A: gtk-1.2.8. including development packages, same goes for the rest of the packages listed. gtk+extra-0.99.12. (Optional) gnome-libs-1.2.8. python-1.5.2. It is unknown whether 2.0 works or not. By all accounts 2.0 should work fine. python-numpy, also called Numeric or Numeric Python, version 1.11 is known to work. Please send reports on success/failure with other versions. libreadline>=4.0. There is a bug in older versions that allows the user to press Ctrl-C, and kill the terminal. 1.2 I can't compile Scigraphica. Configure ran fine, but I'm getting the following errors: Entering directory /home/user/tst/sg-0.6/zvt' gcc -DHAVE_CONFIG_H -I. -I. -I.. -I.. -I./.. -I../intl -I../intl -I/usr/local/include -I/usr/lib/glib/include -I/usr/X11R6/include -DGNOMESBINDIR=\""/usr/local/sbin"\" -g -O2 -I/usr/lib/glib/include -I/usr/X11R6/include -I/usr/include -DNEED_GNOMESUPPORT_H -I/usr/lib/gnome-libs/include -I/usr/lib/glib/include -I/usr/X11R6/include -DWITH_GNOME -c subshell.c subshell.c: In function ^Receive_fd': subshell.c:230: storage size of ^Recvfd' isn't known subshell.c:232: _RECVFD' undeclared (first use in this function) subshell.c:232: (Each undeclared identifier is reported only once subshell.c:232: for each function it appears in.) make[2]: *** [subshell.o] Error 1 A: You need to install all the packages listed under 1.1. Really, we mean that. Leaving out one of them is not an option. Especially leaving out gtk, gtk+extra, python or numeric python will make the compile fail. Seriously though, we know the configure process should pick up that one of these are not installed, it is especially flakey w.r.t. numeric python. If you have a better test for numeric python, PLEASE help us out. 1.3 My version of numeric python is installed under a different directory than the rest of my python installation. How can I make the compile process pick this up? A: Try setting the PYTHONPATH variable to your personal installation directory (the one named site-packages). E.g. for bash et al.: export PYTHONPATH=/home/user/lib/site-packages 1.4 How do I report compilation failures? A: Post your questions to the scigraphica-devel mailing list, including the output from make, as well as your config.log. This will enable us to get as much information as possible and get to the bottom of the matter quickly. The mailing list address is: sci...@li... You don't have to be subscribed to the list to post messages. But it might be worthwile doing so if you have to compile SciGraphica yourself. 1.5 I've compiled a binary for my distribution. Where can I upload it so everybody can use it and save some time? A: You're a great person, we love you. Really. If there were more of you the world would be a better place. More practically, send us a message on the scigraphica-devel list, and we will gladly put the file up on the sourceforge project download page. Or you can upload the binary to ftp://download.sourceforge.net/incoming and tell us the filename. Please include a concise description of your configuration that we can use as a release note. And do realise that we will direct any questions that we feel is caused by your binary to you. Or we may remove the binary from the download area. This sounds nasty and evil, but you would also like to know that the binaries you download are without problems, wouldn't you? Call it our quality control prerogative. 1.x Why does the FAQ end here? A: Because you haven't sent us your Qs and As yet! -- *-----------------------------------------* | Conrad Steenberg | | Caltech, Mail Code 356-48 | | Pasadena, CA, 91125 | | e-mail: co...@he... | | Tel: (626) 395-8758 | *-----------------------------------------* |