roknovosel

Thank you for the feedback, I agree that the journey is a bit too slow at the moment.

Can you try again? I tried it, and it should work now. It hanged due to a bug in my Letterboxd parsing code.

I'm having some issues accessing Letterboxd at the moment. Working on a fix now.

It's not a pure CoPilot alternative, but I'd put https://codesearch.ai into the mix (disclaimer, it's my side project).

It is a semantic code search tool that can be queried using natural language. It provides decent answers to a variety of questions, and I've been finding myself using it quite often to "autocomplete" various mundane tasks. For example, plotting with matplotlib, making http requests in Go, running multiple goroutines, etc. - things where I would usually reach for Google. It doesn't provide a straightforward ready-to-run answer like CoPilot, but it does provide a way to help yourself. It all depends on what you prefer and how you learn. Arguably, having to read the code before you use it makes it more likely it will stick in your brain.

Great read, thanks for sharing. Would love to see the natural language + code mixed in there :)

I've been interested in contrastive learning for a while, mainly as a means to train semantic code search models. OpenAI released a great paper on this topic called Text and Code Embeddings by Contrastive Pre-Training[1] that outlines the approach. I've used it as a base to build https://codesearch.ai [2] with pretty good results.

[1] https://arxiv.org/pdf/2201.10005.pdf [2] https://sourcegraph.com/notebooks/Tm90ZWJvb2s6MTU1OQ==

I agree that semantic code search would be a nice tool in exploratory and onboarding scenarios. The GitHub blog you posted and the CodeSearchNet project was a catalyst that got me interested in semantic code search. That's why I've been hacking on https://codesearch.ai in my spare time at Sourcegraph, which implements a version of semantic code search w/ transformers.

Yep, I'm planning to open source it soon.

Yep, frontend needs some more attention :) And yes, multiple website sources is the ultimate goal.

Not at the moment, but I'm looking into ways to add it. Just appending the language to the query works well for SO posts, less so for the GH functions.

I've also been experimenting[1] in the natural language <-> code world and the two publications based on Tellina have been really helpful. I know how hard it is to get sensible results when mixing natural language and code, and even though Tellina is a bit rough around the edges it is an awesome effort!

[1] https://codesnippetsearch.net

I can't recommend 'Lepa sela lepo gore' (https://easterneuropeanmovies.com/comedy/pretty-village-pret...) enough. IMO, it is one of the best and most underrated war films of all time. A tense depiction of the Bosnian conflict with a healthy amount of dark comedy thrown in there.

I've been working on a small Django project called https://CodeSnippetSearch.net. It allows you to search through code snippets using natural language. Currently, Python, Java, Go, Php, Javascript, and Ruby programming languages are supported.

As with any cool project nowadays, CodeSnippetSearch is powered by neural networks (six in fact - one for each programming language). The project is open-sourced and you can read about the implementation details here: https://github.com/novoselrok/codesnippetsearch

I've been working on a small project called https://CodeSnippetSearch.net. It allows you to search through code snippets using natural language. Currently, Python, Java, Go, Php, Javascript, and Ruby programming languages are supported. I think it's especially useful for new programmers because it allows you to search through code in a "Google-like" fashion. It also allows you to find similar code snippets to the ones found in search results. This enables you to explore different possible solutions.

As with any cool project nowadays, CodeSnippetSearch is powered by neural networks (six in fact - one for each programming language). The project is open-sourced and you can read about the implementation details here: https://github.com/novoselrok/codesnippetsearch This project started as a reimplementation of the models in the CodeSearchNet challenge by GitHub (https://github.com/github/CodeSearchNet/). I have reused their data and reimplemented the neural bag-of-words model in Keras. I didn't expect any improvements with my reimplementation, but I did manage to beat the baseline models by a little bit.

The search is still a bit of a hit-and-miss and I'm continually trying to improve it. If the match rating for the top result is below 50% it will most likely be irrelevant.

Link to repo: https://github.com/novoselrok/codesnippetsearch

Wow, never thought I'd see my paper[1] on HN :)

Even though getting the right performance was tricky, I was more frustrated with the lack of tooling, sometimes cryptic compiler messages and long compilation times for even mid-sized programs. Most of the issues were mitigated by the great community around Chapel. They are usually very responsive and helpful in their Gitter channel.

In regards to actual parallel programming, Chapel was a breeze. Documentation is solid and most of the parallel concepts were easy to use.

Hey, what kind of things do you dislike about the CodeReview SE and what would you like to see in an improved code review system?

I've had an idea for some time now to create a website that would act as a better codereview.stackexchange.com. It would incorporate some of the features of the GitHub Pull Request system like inline commenting and reactions.

I arrived at this idea from two directions. The first direction is that I sometimes try to code review some of the questions over at CodeReview SE, and the whole thing feels unergonomic. I dislike scrolling up and down to check the code and constantly losing track of the things I'm reviewing. This is where I think inline commenting would help. Also, there is not a lot of room for discussion. You only get those comments below the review, where you only have a few characters to argue your point. The second direction is that I produce code snippets (programming homework, short snippets at work, etc.) which I would like to submit for review. I don't always want to submit it to the entire internet for review. I just want to get a private link to the code review, which I can share with my colleagues so they can review it. Kind of like a reviewable PasteBin.

Some of the features I would like to add: importing files from GitHub for reviewing and users could import their unanswered CodeReview SE questions for another review.

I just hope they remove the awful @threads macro with https://github.com/JuliaLang/julia/pull/22631

Developing anything multithreaded was a major pain.

I've been researching parallel programming languages for my thesis and Julia has been really fun to work with. Alongside Chapel, they really make parallel programs fun to write. Although, in my experience, it's been really hard to make them run as fast as C+OpenMP. Mostly because I got trapped in a lot of performance pitfalls as a newcomer to the language. E.g.: parallel TSP branch-and-bound algorithm where I can't seem to improve the performance - https://github.com/novoselrok/parallel-algorithms/blob/maste...

Ljubljana has already made a similar effort: https://www.citylab.com/design/2016/02/how-ljubljana-turned-...

Ok, I'll test it on Chrome 70. Emscripten Module loading is not still as reliable as I would like :/