Earlier quoted context omitted.
Hey, looking into this now. I believe the vote buttons actually do always send us feedback, but the UI could probably be better there.
Unless you are batching them up and hiding them in the upteen bazillion posthog POST requests, no, they do not I shudder to think how much you must be spending on compute asking for that /api/poll/batch once a second, each one saying it is a Miss from cloudfront. It's especially mysterious since there's nothing on the UI that the user would see about the .filesProcessed increasing that justifies DDoSing yourself like…
Launch HN: Greptile (YC W24) - RAG on codebases that actually works
91–100 of 173 posts
Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#92Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#93“Where we going we don’t need docs”. That scares me… docs are among other things there to provide context and info for things not clear from why certain choices were made or not made… no way your AI is going to guess that I put that restriction because of an explicit request from product, despite it looking wrong…
This sort of tech will just spur developers to put proper comments in their code, which they should have been doing in the first place.
Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#94“Where we going we don’t need docs”. That scares me… docs are among other things there to provide context and info for things not clear from why certain choices were made or not made… no way your AI is going to guess that I put that restriction because of an explicit request from product, despite it looking wrong…
Indeed. Code describes how something is done, and possibly what is done, but it seldom fully describes why something is done, or why code exists in the first place. The latter is typically the realm of requirements, design documentation, and possibly test plans. Generating such documentation from code seems quite impossible, or at least wrong. It is unfortunate that most open source software is lacking in such docume…
Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#95Ran it on a "real" OSS project of mine ( https://github.com/dvx/lofi/ ), and it was stuck at 99% loading for about 30 minutes. Then, when it finally parsed the codebase, when asked anything it always returns "Error: Internal error while locating sources." Specifically, I wanted to see if it can context switch between TypeScript (used for the front-end), ObjectiveC (used for a few Mac features), C++ (used for Windows…
Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#96Earlier quoted context omitted.
We’ve been grappling with making it local but two considerations 1. Haven’t figured out what should trigger updates. Every commit sounds crazy, every save would be perfect but too intensive also. 2. LLMs will not be self hosted. We haven’t been able to get the same results with any smaller LLMs To answer your questions we definitely want to make this available for local repos and ideally self hosted, on a technical l…
> Haven’t figured out what should trigger updates. What triggers updates when it is not local?
Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#97Ran it on a "real" OSS project of mine ( https://github.com/dvx/lofi/ ), and it was stuck at 99% loading for about 30 minutes. Then, when it finally parsed the codebase, when asked anything it always returns "Error: Internal error while locating sources." Specifically, I wanted to see if it can context switch between TypeScript (used for the front-end), ObjectiveC (used for a few Mac features), C++ (used for Windows…
Hey sorry to hear that, a couple of things: - processing is usually stuck at 99% because when we order the components of the repository by file-directory dependency and ast dependency, there are a lot, LOT more leaves than the internal nodes + the root. Since we have to have the results of the dependency before we move on, moving up the dependency chain with llm calls take a while. This is even more pronounced when n…
Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#98“Where we going we don’t need docs”. That scares me… docs are among other things there to provide context and info for things not clear from why certain choices were made or not made… no way your AI is going to guess that I put that restriction because of an explicit request from product, despite it looking wrong…
Indeed. Code describes how something is done, and possibly what is done, but it seldom fully describes why something is done, or why code exists in the first place. The latter is typically the realm of requirements, design documentation, and possibly test plans. Generating such documentation from code seems quite impossible, or at least wrong. It is unfortunate that most open source software is lacking in such docume…
You forgot one of the biggest spots for "why" documentation... git commit messages, where the point is to say why you made this change. Maybe taking into account the commit messages around the code in question would help.
Re: Launch HN: Greptile (YC W24) - RAG on codebases that actually works
#99Earlier quoted context omitted.
Indeed. Code describes how something is done, and possibly what is done, but it seldom fully describes why something is done, or why code exists in the first place. The latter is typically the realm of requirements, design documentation, and possibly test plans. Generating such documentation from code seems quite impossible, or at least wrong. It is unfortunate that most open source software is lacking in such docume…
> The latter is typically the realm of requirements, design documentation, and possibly test plans. You forgot one of the biggest spots for "why" documentation... git commit messages, where the point is to say why you made this change. Maybe taking into account the commit messages around the code in question would help.
Still a great idea, though. Ingesting all data from JIRA may be beneficial as well.