Julia 1.0 compatibility

rob-luke commented 7 years ago

Ensure code runs on Julia 1.0

behinger commented 4 years ago

hey! Any plans to continue developing this package? I'm thinking of porting a M/EEG analysis/statistics toolbox for regression ERPs and the likes together. I thought the datastructures you used might be useful :) Best, Bene

Marco-Congedo commented 4 years ago

@behinger Hi, I am also interested in contributing, but it looks it is dead here. Maybe someone should take it over ? See also https://github.com/JuliaNeuro Let us write en e-mail there to see if someone respond.

Tokazama commented 4 years ago

I haven't seen a lot going on at JuliaNeuro for a while. I started https://github.com/JuliaNeuroscience to get things started. I've almost got the beginnings of a core package for properties based on the BIDS standard. If anyone's interested in working on it over there I'd be happy to facilitate.

rob-luke commented 4 years ago

Yes I haven’t worked on this lately. But I would be keen to assist with your new packages. This package was started back with an early version of Julia, so I would do many things differently with modern Julia.

rob-luke commented 4 years ago

But I do plan at some point to make this code Julia1.x compatible. I think I almost had it working a while back.

Marco-Congedo commented 4 years ago

That's great, it would be great to have a BIDS IO package. I can contribute packages for EEG inverse solutions, spatial filters, blind source separation, etc.

abcsds commented 3 years ago

Hi @rob-luke!

I'd like to take over the conversion to 1.6. I'm currently looking for an alternative for my research and I think working on this package is better than making everything by myself. Could we maybe have a chat one day, so you can tell me your design patterns before I make a PR? Thanks!

Best, Alberto

rob-luke commented 3 years ago

Hi Alberto (@abcsds),

Lovely to meet you. I would love for you to take over the conversion to 1.6. There are many improvements I would like to make to this package, but I think its best to start with conversion to 1.6 like you suggest and ensuring the docs and continuous integration etc work first.

To provide you some contex and history on this project. I wrote this package during my PhD and while I was learning Julia, but I had plenty of EEG experience. For the last few years I have mainly been working with fNIRS neuroimaging and python and not EEG, so this package has not been getting the love it needs. However, I have recently been getting back in to Julia (e.g. AuditoryStimuli.jl, DSP.jl, SampledSignals.jl, etc), so would love to get this package revived too.

I live in Australia and am happy to meet from 7am to 10pm, can you find a time that suits between our timezones and send me a meeting request at Robert.luke@mq.edu.au?

Thanks, Rob

abcsds commented 3 years ago

I wrote to you! Thanks for the contact. I look forward to it :)

deep-introspection commented 3 years ago

@rob-luke @abcsds Happy to take part in the joint effort to develop the library too!

likanzhan commented 3 years ago

Are there updates on this project?

rob-luke commented 3 years ago

@abcsds emailed me and I totally dropped the ball. My apologies everyone.

How about we have a call 5pm AEST this Wednesday 11th August to discuss? Would this work for @deep-introspection @abcsds?

deep-introspection commented 3 years ago

Would have love to join but that's 3am in Montreal, so a bit too early...

rob-luke commented 3 years ago

Can you suggest an alternative time that would suit you and @abcsds in Austria too? Im available 8am to 10pm (but can be flexible due to multi time zones, but I agree 3am is too much!).

deep-introspection commented 3 years ago

8am Montreal would make 10pm Melbourne, and 1pm Austria. Would it work for you?

rob-luke commented 3 years ago

Perfect. Which also makes me free Monday, Tuesday, Wednesday. Any of these suit you?

deep-introspection commented 3 years ago

Both Tuesday and Wednesday could work for me. @abcsds ?

rob-luke commented 3 years ago

Ok I have made a meeting for 10pm AEST on Tuesday 17/8/2021

Join at https://macquarie.zoom.us/j/83514916532 Password is Julia

Let me know if you want to move to Wednesday @abcsds

abcsds commented 3 years ago

I can make it tomorrow at 10pm AEST. I might be 5 min late. See you then!

likanzhan commented 3 years ago

I will also like to join in, if my daughter is not too noisy.

rob-luke commented 3 years ago

Awesome, that's why I posted it publicly, the more the merrier.

abcsds commented 3 years ago

Maybe I'm too late, but the password is not working for me.

rob-luke commented 3 years ago

I think you are 5 hours early

deep-introspection commented 3 years ago

@rob-luke Are you there? Can’t connect with “Julia” either.

likanzhan commented 3 years ago

me neither

rob-luke commented 3 years ago

yeah

rob-luke commented 3 years ago

Ill make another meeting

rob-luke commented 3 years ago

https://macquarie.zoom.us/j/83514916532?pwd=NGYvM1V5bDNmOFF1K2NVa3pSWGFidz09

rob-luke commented 3 years ago

Thank you all for joining last night. For anyone interested who didn't join the plan is roughly:

Build on this package rather than start a new one. As that will be less work than starting fresh.
- Step one is to get continuous integration up and running, to enable faster code reviews.
- Step two is to upgrade the code two Julia 1 compatibility.
- Considering jumping to 0.7 then 1.0 as there might be better user warnings to ease transition
- Look for tools to help with transition
- Step three is people can add whatever features they wish. Some reworking of the type structure was proposed and addition of connectivity metrics. I am open to these suggestions.
My personal preference is to get things working robustly with tests before optimising
My personal preference is to add most used features first. E.g. basic signal processing used in every study before adding complex graph based analysis/processing and multi threading support etc.

Help is appreciated with all aspects of this plan.

Tokazama commented 3 years ago

Sorry I couldn't make the call. I'd recommend getting technical tests working and put API and performance specific stuff off for a bit. In order for us to have strong and maintainable performance we should harmonize with what the compiler experts are moving towards. Then design API around that, allowing us to inherit computational efficiency by design.

rob-luke commented 3 years ago

Thanks @Tokazama, I welcome any PRs. But I will first be getting the existing API running as it was very functional. Then expanding. Please open specific issues if you find any places in the code that could be improved by design. Improvements are welcome.

rob-luke commented 3 years ago

Done. But now comes the hard part 😆

rob-luke commented 3 years ago

I am raising issues for all problems I encounter. I am marking ones that I think are nicely self contained and good for new contributors as "help wanted". Feel free to jump in to any of these @abcsds (or more complex ones if you want :wink:)

rob-luke commented 2 years ago

Ok everyone, the existing code is now reasonably documented. Feel free to open PRs to improve any aspect of the codebase. Thanks for your patience

rob-luke / Neuroimaging.jl

Julia 1.0 compatibility #21