marceloabsousa

Hey HN, Reviewpad co-founder here. Happy to answer any questions about this launch - excited to hear your feedback about it!

I find this clearly bogus.

I wouldn't discard the capabilities of the LLM. I've been using it to do relational reasoning of Java comparators and it is surprisingly accurate. https://twitter.com/marceloabsousa/status/163289383288582963...

This is harder than it seems. The way we are doing with Robin (Reviewpad's AI Reviewer) is by feeding it context about the PR and results from our static analysers and allowing the developer to prompt directly through the PR comment. For example, https://github.com/marcelosousa/robin-preview/pull/2

At Reviewpad we're also working on a similar product.

Just yesterday, I've tweeted about it: https://twitter.com/marceloabsousa/status/163289383288582963....

Some pull request examples at: https://github.com/marcelosousa/robin-preview/pulls

It all depends on what you mean by generate fully functional code. In my experience, it is able to generate code snippets that are useful but very rarely can be simply copy & pasted into your work. The experience is actually quite similar to Stack Overflow.

Maybe I’m missing something but isn’t the whole point to actually release api.py? Everything else seems trivial.

I have started my startup at 30 when I was engaged (now married).

Every path is different but I think there are a couple of guidelines that I recommend: - Talk with your partner and set expectations from the beginning. Please do not romanticise the journey - it's damn hard and full of sacrifices. Others have suggested a deadline - I also did that but find it a bit unrealistic because you'll have excuses to continue devoting an unhealthy amount of time to your company. Be kind to your partner and acknowledge that they will take more load in family duties and you'll have less time to spend with them.

- Find a co-founder that is complementary to you and that you trust. Being a solo founder in your 30s with a family is super hardcore even if you already have found product market fit.

- Do not invest your own money. Finding an early champion of your project and your founder team that could be an angel investor to kickstart things is a must. The alternative is to do it as a side gig until you can reach some kind of Ramen profitability scenario - I'm not a big fan of this because you'll likely focus on revenue too early on.

Happy to chat about my experience if you think it'll be valuable (just DM)

I think that there's been a lot of research about this. I can recommend the book "Change: How to Make Big Things Happen" by Damon Centola. Ironically, it goes into detail on how Twitter became successful.

Thank you for this post. It’s nice to see that it wasn’t buried. From the times of IRC, I do miss the feelings of the early times of the internet.

I’ve the privilege to invest a lot of time thinking about what I should be doing my life professionally. In the end, the best I could think of was to be an enabler for others - which for me meant ‘to give them time’. I realised that I also needed to use my expertise to try and give this purpose a better chance of success. I have a PhD in static analysis and programming languages and got experience with writing code analysers.

So I’m on this quest to combine these two core ‘ideas’ - use static analysis to help developers save time and hope that they can improve their life with that time.

I managed to start a company around this mission and build a product focused on reducing the time on pull requests. it’s been a very hard journey so far and feels like I’m just getting started still. It makes me sad and sometimes lonely to see that we are ever more connected but that many times it feels that no one really cares. In my early days of using the internet it was much easier to make meaningful connections.

I find mastodon a bit hard to navigate between the servers and how to engage in a particular server.

Second this. After working fully remote for 2+ years and now back to the office, for me it's really great to have an office.

One of the languages I've designed is called SLANG (https://github.com/sonarsource/slang) when I was working at Sonar.

The ideia is to have a common IR representation from multiple input programming languages to perform static analysis on top of.

This is useful because it's very expensive to write dedicated backend analysers when you have to support 25+ programming languages.

This is specially true when analysers share a subset of the checks (e.g. cyclomatic complexity, identical conditions check). You can check the full list at: https://github.com/SonarSource/slang/tree/master/slang-check....

Author of the post here! Happy to answer any questions.

TLDR: Not all pull requests are the same!

We are working on a way to specify and automate PR workflows so that you can separate the code review and the PR approval/merge steps.

This way you can automate which PRs require code reviews and which don't.

We show how we are using Reviewpad (as a GitHub Action) to automatically merge basic changes and also how to use it to ship PRs ensuring the CI passed.

Thanks! Feel free to join our discord at https://reviewpad.com/discord to give us some feedback!

This is a classic case of being overwhelmed by complexity, when in truth complexity is an advantage. SaaS services tend to be your biggest friend on the first month of usage, and them when you need more that's it, you're stuck to that.

This article more about parsing at scale than static analysis at scale.

If you don't want just a tree but also a semantic diff that tells you what types and methods where actually changed check out Reviewpad (http://reviewpad.com).

Since software developers (and then engineering managers) get outrageous salaries it has been easier for some of them to save up substantial amounts of money that allows them to try and build something from scratch.

I'm improving the state of code reviews with Reviewpad. It's been an hard journey.

For large PRs, I ask the PR author to perform a 'guide' review and then use Reviewpad (http://reviewpad.com/) to be able to nicely iterate over the comments of this guide. This way, I don't need to jump one a call or use loom but I can understand the code changes.

We have been on working on this for code review purposes at reviewpad.com. Would love to catch up and see your use case!

Totally agree. Co-pilot is just stressing more our roles as code reviewers. It’s not even a new idea - people working in program synthesis have done this for a while now. I’ve written about this a couple of years ago: https://medium.com/@marceloabsousa/the-software-shift-toward...