-
Notifications
You must be signed in to change notification settings - Fork 26
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
curious doctest failure #685
Comments
Threading has a lot to answer for, especially in transient failures. In this case, I would be most interested to figure why one has threads and not the other. |
For info,
Gentoo:
|
Very strange. I was expecting some other libraries to be involved. |
There are other Python symbols that are undefined - I'm assuming they're picked up at runtime? Vanilla:
Gentoo:
|
I think the linking to pthread is bogus and comes from not using |
However when I build vanilla I have exported
|
And yes, python symbols are provided by python when it loads the module. It would be an issue if something other than python (or cpython more precisely) was loading them, which is an issue with some gap libraries. |
I have run out of ideas there. The next step would be linking flags coming from python. But both are using the same python as far as I can see. It also suggests that the Gentoo version is underlinked with regards to pthreads, but it would show in the output from Would have to inspect the linking line in the log for the vanilla version. I can use my own copy for Gentoo. |
I must say the build log is not very informative. All I have is
I can't find any linking info. |
Darn, same as gentoo log. We have to find a way to restore verbosity. |
That's funny, I get the failure when I do testing from inside the ebuild (something I am currently working on thanks to new bits of distutils-r1 see 1468853) and therefore test on the original source file but not when I do a separate doctest with the installed sage. |
I do have an old s-o-g build log from about a year ago (May 2021) in Prefix
The build failed for some reason. |
Perhaps this would be easier to debug since there is only one |
A poor mans
On vanilla I did a |
Yes, I am not convinced at all that |
I have verbosity.
in the function |
Good news. I also added
with no change to |
Really not sure where it come from and whether it is really missing in s-o-g. I am having random compile error on my machine these days :( the pattern makes me suspect faulty RAM. I'll have to figure how to test it. |
Well, I get verbose output on vanilla to a point until in building
The
and
The |
The [351/531] should not be in the above post. The subsequent line is OK. |
Trying again, I had |
Success! The above compile/linking is correct for vanilla. However, I see no
There is no
|
Perhaps the |
The |
Vanilla
and
but results similar to Gentoo output without the
and
Not sure what is controlling this different output. The only new |
I just merged the vbraun branch into master. s-o-g is verbose from now on. Anyone with |
For the record verbosity is off again. My current patch appears to be useless now. Not sure what caused verbosity to switch off. |
Do you know whether this started with |
No this happened earlier, I just didn't stop to put a note about it. I may have noticed around 9.7.beta1. |
|
Let me check what I do with sage_setup in the sage ebuild these days... |
Indeed, for a while I was moving sage_setup out of the way so it wouldn't be used. But it caused issue with testing. Good spotting. I will just chuck it out as I have decided on a different testing strategy for now. |
On vanilla (9.6.beta5) I have
which is resolved by https://trac.sagemath.org/ticket/33496. However, the doctest passes on s-o-g as is. Is my machine
slower
when running s-o-g or something else? I did notice that the vanillacoerce_actions.cpython-310-x86_64-linux-gnu.so
was built with threads. The s-o-g build does not use threads.The text was updated successfully, but these errors were encountered: