Why do so many tests fail on a clean Sage installation

doctesting

asked 2010-09-16 11:46:39 +0100

With a completely clean Sage installation, performing ./sage -testall as suggested (required) in the "Walking Through the Development Process", many doctests fail with various errors (including a few segfaults). Is this normal? If so, I'm sure that it is not intended, so is there someone with a grand overview on whether these errors are all in the process of being handled?

These errors mess up the output and makes it very difficult to sort out which are errors caused by the patch you're working on, so you might miss them. It's like dirty dishes in a shared kitchen: if there are _any_, these will quickly multiply ;-) Is there a way to suppress those errors known (assuming the errors are not due to something I did wrong), so I can get a clean overview of new errors?

Cheers, Johan

edit retag flag offensive close merge delete

add a comment

2 Answers

Sort by » oldest newest most voted

answered 2010-09-16 11:59:22 +0100

kcrisman
12242 ●42 ●136 ●255

This may depend on your machine. I don't think this "answers" your question, but these things are really dependent on the situation. In general, a fairly modern (i.e. fast) machine and normal distribution should not have any doctest failures (other than optional ones) and only a couple timeouts. (I get lots of timeouts on some of my machines, but they are also >5 years old, and setting SAGE_TIMEOUT or SAGE_TIMEOUT_LONG usually fixes that.)

So I guess what I'm saying is that it's better to first make sure your patch is removed, that no extra spkgs are included, run the doctests again and then report them immediately either to sage-release or sage-devel. Even if you have a non-standard setup, our goal is to fix any bugs we can find. Please let us know!

If you are using a prerelease version, on the other hand, we usually expect a certain number of errors to arise as people test them on various platforms.

edit flag offensive delete link

Comments

I have a brand new thinkpad t410 with quadcore i7 processor, running an up-to-date Ubuntu, so that should be ok. I'll try to make sure I have a "stable" Sage and redo the tests.

jsrn ( 2010-09-17 03:37:48 +0100 )edit

Ok, now I downloaded the full source, compiled it myself and ran all the test, and this looks much better: only two errors (I'll post on sage.devel). I think my old version was the pre-compiled 4.5 for Ubuntu 64 bit...

jsrn ( 2010-09-18 07:34:27 +0100 )edit

add a comment

answered 2010-09-16 15:22:49 +0100

niles
4529 ●23 ●72 ●133 http://nilesjohnson.net/

A good installation of sage should have no errors, but this problem has come up for me too :(

If you add information about your system, perhaps someone here can help you. If not, you could try re-posting your question (with system details) on the sage-devel list.

In the meantime, if sage is working you can still continue development and test sage on the subset of files you think will be affected by your changes--this makes keeping track of new doctest failures a little easier. Of course you will need to test all of sage eventually, but you don't have to just do noting until this gets resolved :)

edit flag offensive delete link

add a comment

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Why do so many tests fail on a clean Sage installation

2 Answers

Comments

Your Answer

Question Tools

Stats

Related questions

Why do so many tests fail on a clean Sage installation edit

2 Answers

Comments

Your Answer

Question Tools

Stats

Related questions

Why do so many tests fail on a clean Sage installation