`numpy.arange()` / `numpy.sum()` segfaults

[plajjan]

examples/sumto2 seems to crash with a segfault. Would be good if we could create a proper test case instead of just an examples. We do test to make sure that all examples compile but we don't actually run them (since we don't know what the expected outcome should be, some example apps are interactive, others are not, some expect arguments etc).

@sydow will you pick up on this one?

[sydow]

Hi,

I could have a look and fix it, but now I am working to profit from the improved rows to get better error messages. And after that I would like to try to fix list/set/dict comprehensions. So probably not this week.

And also: should we keep the numpy stdlib at all? I think I am the only one who has an interest, and what we have is such a small part of the Python numpy functionality that it is not obvious that it is worthwhile to keep it.

Björn

On 8 Jan 2024, at 15:15, Kristian Larsson @.***> wrote:

examples/sumto2 seems to crash with a segfault. Would be good if we could create a proper test case instead of just an examples. We do test to make sure that all examples compile but we don't actually run them (since we don't know what the expected outcome should be, some example apps are interactive, others are not, some expect arguments etc). @sydow will you pick up on this one? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>

[plajjan]

@sydow this issue isn't particularly prioritized right now, I just noticed it when I tried to change arange() to accept the same type of arguments as range() as per the conversation in #102. I think the improved error messages following improved rows and comprehensions are much more valuable and indeed higher priority! Looking forward to your work there :)

I can't really say if we should ditch numpy or not. It's not really used but it doesn't seem to be that much extra work, right? We could of course rip it out and reintroduce at some later point in time. I don't have any strong arguments for or against it :shrug: