Frequently Asked Questions About Sage Development

I'm a beginner to Sage development. What can I do to help develop Sage?

See the guide How to Begin Sage Development.

I'm not a programmer. Is there another way I can help out?

Yes. Like any free open source software project, there are numerous ways in which you could help out with developing Sage. And programming is one of many ways to contribute. If you can speak, read, and write in another (natural) language, there are many ways you can help out. Say you know Italian. Then you can write a Sage tutorial in Italian, or help out with translating the official Sage tutorial to Italian. For the graphics designers or artistically creative, you can help out with improving the design of the Sage website. Or you can cast your critical artistic eyes over the Sage notebook interface and find out where it needs improvement.

Many people like writing technical tutorials. One of the joys of doing so is that you also learn something new in the process. At the same time, you communicate your knowledge to beginners, a skill which is useful in fields other than technical writing. A main point about technical writing is that you communicate a technical subject to beginners, so keep technical jargons to a minimum. Darrell Anderson has written some tips on technical writing, which you are highly recommended to read

The above is a very short list. I’m sure there are many, many more ways in which you can help out.

Where can I find resources on Python or Cython?

See the wiki page on Python/Cython resources.

I want to develop code using functional programming. Where can I find resources on Python functional programming?

See the Sage Constructions document for tips on Python functional programming. These tips are organized under the chapter Python Functional Programming for Mathematicians.

Are there any coding conventions I need to follow?

You should follow the standard Python conventions as documented at PEP 0008 and PEP 0257. Also consult the Sage Developers' Guide, especially the chapter Conventions for Coding in Sage.

I have written some code for the Sage library. How do I test the whole Sage library with my new code?

See the chapter Parallel Testing the Sage Library of the Sage Developers' Guide.

Are there any guidelines on using the Sage bug server?

See the chapter The Sage Trac Server: Submitting Patches and Packages in the Developers' Guide.

A ticket provides a replacement spkg for reviewing. How do I integrate such a replacement spkg into the latest development version of Sage in order to review the ticket?

The details of the review are slightly different depending on whether the replacement spkg is:

1. a standard spkg;
2. an optional spkg; or
3. an experimental spkg.

Standard spkg's live in the directory

SAGE_ROOT/spkg/standard

First, ensure that you have access to the latest development version of Sage. To review a replacement standard spkg, you could integrate it into the latest development version installing the spkg. The relevant command is

./sage -f ULR/to/spkg

This would download the spkg, compile it, and install it---all done automatically for you. Once all these steps are done (i.e. the spkg has been integrated into your copy of Sage for reviewing), you should rebuild the Sage library (in case the new spkg also affects the Sage library):

./sage -b main

Now you are good to review that the replacement spkg fixes any reported problems.

Another way to integrate a replacement standard spkg is to take a source tarball of the latest development version of Sage, uncompress it, remove the current version of the spkg as found under

SAGE_ROOT/spkg/standard

and drop the replacement spkg under that directory. After that, issue make from SAGE_ROOT to compile the whole Sage distribution. This will compile everything, including the replacement spkg you have placed under

SAGE_ROOT/spkg/standard

This is usually a safe way to ensure that the replacement spkg also plays nicely with the other standard spkg's and the Sage library. It also ensures that everything still compiles from source without any errors.

Integrating optional and experimental spkg's work similarly as described above. However, you only need to install the replacement spkg as follows:

./sage -f URL/to/spkg

So it can take more work to integrate a standard spkg than it does to integrate an optional or experimental spkg. The extra effort is required since a replacement standard spkg must not break anything when integrated into Sage.

When I rebuild the Sage library, can I save everything under SAGE_ROOT to a backup directory and later on copy it back when I want to restore to the original after a patch?

Yes you can. Just to clarify, you could do this: You could copy your Sage directory to somewhere else with something like

cp -r sage-x.y.z some/where/else

Then you could proceed with playing around/reviewing a ticket on top of your Sage library as contained under sage-x.y.z/devel/sage-main. (But you need to first apply the patch.) When you are done with sage-x.y.z, you could trash it if you want. Now you still have that pristine copy of Sage you have copied to

some/where/else

Let's call that copied version sage-copy. Now sage-copy is not in a working state because when you copied, everything from sage-x.y.z (including binaries and path variables) were copied over to sage-copy. To get sage-copy into a working state you could use, you need to cd to its SAGE_ROOT and regenerate lots of stuff, including path variables, as follows:

./sage

What you could also do is make a clone of your working Sage version, which is significantly faster and takes up much less space on your hard drive. Say you start with sage-x.y.z, which can be a binary version you've downloaded, or you have compiled it from source. To make a clone of the Sage library you could experiment with, cd to SAGE_ROOT and do

./sage -clone <name-of-clone>

This would create a working copy of the Sage library and named

SAGE_ROOT/devel/sage-<name-of-clone>

The alias

SAGE_ROOT/devel/sage

always points to the current branch of the Sage library you're using. The main branch is called

SAGE_ROOT/devel/sage-main

To switch from the main branch to your cloned branch, do

./sage -b <name-of-clone>

For example, say you built Sage from source and SAGE_ROOT is ~/sage_current. You then made a backup copy with

cp -r ~/sage_current ~/sage_backup

and you applied patches or install new spkg's to ~/sage_current in order to review those patches or spkg's. After you're done reviewing, you delete your copy of Sage, and restore the backup copy, with

rm -r ~/sage_current
cp -r ~/sage_backup ~/sage_current

This would still give you a working copy of Sage. Both ~/sage_backup and the newer ~/sage_current, copied from ~/sage_backup, are still working copies of Sage that you originally copied from the original ~/sage_current. Now you could do

cd ~/sage_current  # or cd ~/sage_backup
./sage

and you'd be presented with a working terminal session of Sage. What could also do is

cd ~/sage_current  # or cd ~/sage_backup
./sage -b

Warning: This would regenerate/re-compile files in the Sage library to reflect the (possibly) new paths of these copies. The regeneration/re-compilation takes a long time even on a high-end server like a machine on the Sage cluster.

Apart from the above strategy for making backups of your working Sage installation, you could also consider making your own binary distribution. The steps are simple. First, take a source tarball and compile Sage from source. Once it's compiled, do

./sage -bdist x.y.z  # or some name other than x.y.z

to get a binary distribution. Once the binary distribution is all packed up for you, the binary tarball could be found under

SAGE_ROOT/dist

You could take that binary tarball as your backup copy. Every time you want to experiment with new code, uncompress that binary tarball, and play away. Note that the binary tarball contains the source of the Sage library and binary versions of standard spkg's. Unlike the source tarball, the binary tarball doesn't have the source of standard spkg's. For more information on Sage command line arguments, do

./sage -advanced

This would list all command line options you could pass to Sage.