Make numpy an optional dependency about imageio

[FirefoxMetzger]

I am closing this issue due to inactivity (for almost 8 years ⚰️ ) I think by now numpy and, in particular, the array format is such a widely supported API that having a numpy dependency is more of a strength than a hindrance.

If, for any reason, a numpy dependency still poses a problem feel free to leave a comment and we can re-open this issue or make a new one, depending on the type of issue.

[almarklein]

Coming from the scientific community, numpy is not such a bad dependency, since almost every library that we use depends on it.

But I see your point; in order for imageio to be useful in e.g. the web community, the numpy dependency is an obstacle.

Since we mainly use numpy as a container for the data and don't do any processing, I suspect that making the dependency optional should be relatively easy.

[sametmax]

Cool.

I must say I have been waiting for a project like this for a long time.
Right now, you can make any Web Python code dealing with image
processing newcomers friendly, and they complain very hard given how
easy it it to do the same on a PHP project. This, and cryptography, are
the two batteries Python has not included.

Good luck.

Le 17/11/2014 16:00, Almar Klein a écrit :

Coming from the scientific community, numpy is not such a bad
dependency, since almost every library that we use depends on it.

But I see your point; in order for imageio to be useful in e.g. the
web community, the numpy dependency is an obstacle.

Since we mainly use numpy as a container for the data and don't do any
processing, I suspect that making the dependency optional should be
relatively easy.

—
Reply to this email directly or view it on GitHub
#41 (comment).

[ghisvail]

Since we mainly use numpy as a container for the data and don't do any processing, I suspect that making the dependency optional should be relatively easy.

Technically, you could probably replace the numpy.ndarray container with a array.array for the data However, I don't know about things like multidimensional indexing and conversions between dtype which numpy definitely makes easier.

[almarklein]

I'm working on https://github.com/almarklein/tinynumpy for this