Open mikolalysenko opened 9 years ago
No worries! Unfortunately, I'll be communing with nature until June 9, but then I'll be back at it! And able to do some research/ organizational work in the mean time.
A couple details on my stuff: The qr factorization module handles overdetermined least squares problems, but SVD + eigenvalues will be a bit more work. My primary reference so far just gives general direction on eig/SVD, and I miss having academic journal access, so I have a bit of research to do.
Also, I think it'd be a good idea to create some sort of generator for these repos so we can move toward some sort of consistency.
Side note: an ndarray repo showed up in compute.io so I feel like we're still duplicating a bit of effort there. I realize they're two distinct projects, but it's not clear to me that the goals are so different—and there's certainly enough work for as many as are interested—that it would interfere with anyone's vision to try to share effort.
Sorry, not to drive that point too hard, but just trying to think about the big picture stuff so that the result doesn't end up off in one way or another. Some of my work is currently a bit inconsistent at the moment.
Despite all the caveats though, I remain excited, enthusiastic, and available on June 0 to get back to it!
Ricky Reusser
On May 27, 2015, at 3:30 PM, Mikola Lysenko notifications@github.com wrote:
Starting next month I want to work on getting SciJS into a publishable state. The following things need to get done:
Marketing
Website Logo Documentation
Quick start guide Documentation registry (similar to http://stack.gl/packages ) Need to organize modules into categories Contributing guide Tutorials Make TODO list and roadmap more accessible Technical
Consolidate redundant packages Bug fixes, need to harden up cwise More testing across whole ecosystem Update old dependencies and packages Cool demos for specific modules! Future stuff
Sparse operations aren't a priority for this release, but once we have a basic linear algebra ecosystem up and running for dense stuff we can revisit them. Statistical processing would also be nice. With the tools we have here it should be pretty straightforward. Maybe try to publish or present some of this stuff? Need to figure out the right venue People
@hughsk if I create some content for a webpage, could you try working some design magic to make it look nice?
@rreusser how far off do you think you are from getting a working SVD implementation, or even something for just solving overconstrained linear least squares problems? I think that with that we would be able to do a lot of fun stuff, especially with image processing/physics simulation.
@jaspervdg can you maybe take a look at some of the outstanding issues on cwise? I know your defense is coming up soon, but I think that some of them are blocking.
@substack also if you get a moment, can you try to figure out what is going on in static-module that is causing breakages? @rreusser you mentioned that you had a failing case
Apologies
I'm sorry that for the last couple of weeks I've not been pulling my weight and maybe haven't been as responsive as I need to be. I really value all of the hard work that you guys have put in on the project.
— Reply to this email directly or view it on GitHub.
@mikolalysenko I'd be very happy to do so! When're you aiming to have this all ready by? I too am gonna be pretty caught up for a little while, until around June 7.
What're your thoughts on sharing a similar aesthetic between scijs and stackgl? Also, if you're after a documentation registry similar to stackgl's, maybe we should create something reusable – I know that projects like glo would benefit from that too.
@mikolalysenko Currently at a conference, but will take a look next week.
http://scijs.net/packages looks great! Just a small note about assets since my images are broken:
I used the relative path shorthand for images, e.g. ![…](docs/images/example.png)
. Quick and easy but breaks when the READMEs get moved. I assume the preferred method (correct me if I'm wrong) is to hard-code: https://raw.github.com/username/reponame/master/…/imagename.png
. There should probably be a note about this on the packages page.
Need to confirm whether this works for SVGs. (i.e. think through the security implications of the raw subdomain since injection vulnerability seems to be why the shorthand notation doesn't work for SVGs.)
At any rate, there's probably a single path of least resistance that's maybe worth mentioning.
Otherwise great though!
I wouldn't say using relative path shorthand is the wrong thing. I've actually been changing a few of those hardcoded urls into relative paths (also using ![]() instead of img tags), since it is more idiomatic, more readable, and (at least in theory) more future-proof. Most likely we just want to change some scripts so they can transform relative paths (the npm registry appears to handle this just fine).
Ah, there it is: marky-markdown, "The thing npm uses to clean up READMEs and other markdown files". lib/cdn.js rewrites image urls. Nothing fancy; just curious if there was already a module that handles some corner cases.
Not sure the current state of code/interest/other, but the work I did that was kinda okay but kinda loose has been driving me crazy, so I'm gonna revisit what I've done and tighten things up. I've updated the packages documentation to the latest stackgl has to offer. See: http://scijs.net/packages/
I think there's too much good work here to let it drop, so please let me know if there's anything I can do to keep pushing things forward, otherwise I'll do my best to keep chipping away at cleanup, robustness, presentation, etc.
Starting next month I want to work on getting SciJS into a publishable state. The following things need to get done:
Marketing
Documentation
Technical
Future stuff
People
@hughsk if I create some content for a webpage, could you try working some design magic to make it look nice?
@rreusser how far off do you think you are from getting a working SVD implementation, or even something for just solving overconstrained linear least squares problems? I think that with that we would be able to do a lot of fun stuff, especially with image processing/physics simulation.
@jaspervdg can you maybe take a look at some of the outstanding issues on cwise? I know your defense is coming up soon, but I think that some of them are blocking.
@substack also if you get a moment, can you try to figure out what is going on in static-module that is causing breakages? @rreusser you mentioned that you had a failing case
Apologies
I'm sorry that for the last couple of weeks I've not been pulling my weight and maybe haven't been as responsive as I need to be. I really value all of the hard work that you guys have put in on the project.