imageio and imageio-ffmpeg Using isAlive() in Python3.9 #808
Comments
|
Hej @jmdelahanty 👋 A quick note on your setup. According to the exception, the ffmpeg plugin crashes in line 901, which is strange because the current version only has 700-ish lines of code. So I checked your version and noticed that you are still running on ImageIO 2.3.0: imageio 2.3.0 py_1 conda-forge
imageio-ffmpeg 0.4.7 pyhd8ed1ab_0 conda-forgeIs there a specific reason for this? If not, a first step could be to try and upgrade ImageIO to something more recent (at the time of this writing we are on |
|
Oh my I’m very behind then! There’s not a specific reason for that, it’s just what was installed when I was making my environment initially. I guess I should probably upgrade then! I’ll do that and come back and close the issue when the upgrade fixes that problem.
…________________________________
From: Sebastian Wallkötter ***@***.***>
Sent: Monday, May 2, 2022 9:08:11 PM
To: imageio/imageio ***@***.***>
Cc: Jeremy Delahanty ***@***.***>; Mention ***@***.***>
Subject: Re: [imageio/imageio] imageio and imageio-ffmpeg Using isAlive() in Python3.9 (Issue #808)
Hej @jmdelahanty<https://urldefense.com/v3/__https://github.com/jmdelahanty__;!!GX6Nv3_Pjr8b-17qtCok029Ok438DqXQ!3tST9tbeMUeMvvUVh6N5yU_ZUCk2q9UwJmxdjBPmF6ws-LdYBFI0b54HnHcQqzNEAfzksqcW7iZjywwD66gLal5yQjk$> 👋
A quick note on your setup. According to the exception, the ffmpeg plugin crashes in line 901, which is strange because the current version only has 700-ish lines of code. So I checked your version and noticed that you are still running on ImageIO 2.3.0:
imageio 2.3.0 py_1 conda-forge
imageio-ffmpeg 0.4.7 pyhd8ed1ab_0 conda-forge
Is there a specific reason for this? If not, a first step could be to try and upgrade ImageIO to something more recent (at the time of this writing we are on 2.19.0).
—
Reply to this email directly, view it on GitHub<https://urldefense.com/v3/__https://github.com/imageio/imageio/issues/808*issuecomment-1115722827__;Iw!!GX6Nv3_Pjr8b-17qtCok029Ok438DqXQ!3tST9tbeMUeMvvUVh6N5yU_ZUCk2q9UwJmxdjBPmF6ws-LdYBFI0b54HnHcQqzNEAfzksqcW7iZjywwD66gL8hEd2fU$>, or unsubscribe<https://urldefense.com/v3/__https://github.com/notifications/unsubscribe-auth/AGYYKOLAYBVTB4B2BBUR6CDVICRCXANCNFSM5U5PLLQQ__;!!GX6Nv3_Pjr8b-17qtCok029Ok438DqXQ!3tST9tbeMUeMvvUVh6N5yU_ZUCk2q9UwJmxdjBPmF6ws-LdYBFI0b54HnHcQqzNEAfzksqcW7iZjywwD66gLb7rsdBQ$>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
|
@jmdelahanty That sounds like a good start. v2.3 may have been installed because of some dependency; if that is the case, and you happen to figure out which package depends on this older version, please let me know. I will then go and check with the maintainers of that package if there is any bug we are unaware of, or if that version can be bumped. |
|
Doing a simple Do you think that the fact that it's still 10 minor versions behind is a problem at all? I can try and see if there's other packages holding it up, but I'm not really sure how to check... |
If you set up a fresh environment then one or more of your packages have pinned the version to 2.3.0 or below.
In general, you should always try to upgrade to the latest version-major release, unless you have reasons not to. For a specific case, it depends. If you are using this as part of a public-facing service then yes, you should upgrade to avoid bugs and security vulnerabilities; otherwise, it's not too big an issue unless there is a bug that affects your use case.
Unfortunately, I'm not sure if and how conda resolves dependencies since I almost exclusively use pip. I think @anjos might have an idea, since he maintains our conda feedstock. @anjos Do you know if it is possible to visualize dependencies when installing with conda? More specifically, we have a downstream package that has pinned the ImageIO version to |
|
The current run requirements for imageio are defined on its recipe (https://github.com/conda-forge/imageio-feedstock/blob/main/recipe/meta.yaml):
So, if the environment is set in such a way to conflict with those pins, then I guess a lower version will be tried until one is found that seems compatible. The command |
|
@jmdelahanty -- PS.: I find it curious that your environment has a mixture of both conda-forge and defaults packages. IMO, you should avoid this, generally speaking. Use only conda-forge when conda-forge is required. Instead of |
|
Thanks for chiming in @anjos ! So conda does enforce dependencies / does dependency resolution. That's good to know! @jmdelahanty On the same topic, I found a python package that can do a reverse dependency lookup, which is called
@anjos Yes, we no longer need numpy>=1.20 as of Imageio v2.17.0 because I refactored type annotations into separate files and created a fallback in case |
|
Good to know, we should then remove it from the conda-forge recipe. I'll do this later today and issue a new build. This may fix observed issues, or at least relax the environment a bit further. |
|
OK, build 1 is available at the conda-forge channel, with the relaxed numpy pinning. You may want to try again and see if you still hit the same issue before starting to look elsewhere. |
|
Would you like me to try this new build @anjos? I can make a new environment, install the stuff I need, and see what happens? I'll also switch over to mini-forge and try out mamba-forge too! Thank you for the advice! I've heard of mamba/mamba-forge before but have never used it. I can update here what happens, will definitely do it in the next couple of days. |
|
@jmdelahanty: It was just meant as debugging tips, in case you're looking for possible causes of issues. This is not the right thread for debugging conda setups. If you have issues with the conda package itself, I'd advise you open an issue at the repository for the package at https://github.com/conda-forge/imageio-feedstock/. |
|
Got it, thanks @anjos! And for you @FirefoxMetzger, I ran the |
|
@jmdelahanty Okay, in that case, it was probably just an outdated library on your system and nothing major is preventing you from using a more recent version of ImageIO. To sum up all the discussion above, the solution to this issue seems to be "update ImageIO, because the problem is fixed in more recent versions". @jmdelahanty Does that solve your problem? If so, I will go ahead and close this issue to stay organized, otherwise let me know what is missing. |
|
That's all correct! Thank you for your help and teaching me how to check these things. |
Hello ImageIO Devs!
I'm currently using your reader with
PIMSandDasksince it seems likeffmpegbackends are the way things are moving for video editing. It's for this project.While running my code on a Linux machine in an environment that has
Python 3.9.2, I received the following error message:After googling around a little, I found a few other issues that talk about how
isAlivewas deprecated at some point. I'm not really sure why other than it seems like they wanted to move to snake_case instead of camelCase?When I looked at the source code in my Conda environment as the Traceback describes, I indeed found the use of
isAlive! Changing this tois_alivelets things work again!Below is my full environment for reference.
I'm not sure if this is actually more related to Dask or PIMS, but since the traceback came from the
ffmpeg.pymodule I figured I'd post it here.The text was updated successfully, but these errors were encountered: