Live data from Hacker News

Ask Tom Preston-Werner, cofounder of GitHub, anything Today, Mon 18 Oct 2010.

news.ycombinator.com

61–70 of 144 posts

Re: Ask Tom Preston-Werner, cofounder of GitHub, anything Today, Mon 18 Oct 2010.

#65
post #45

How do you decide what features to add or not add to Github? How much thinking, discussion, planning, design, research, etc go into each feature before coding begins?

We look for features that we want to use ourselves. It's usually quite apparent what areas of the site need some love or where entire features are missing. The things that get implemented are the things that are important enough for someone on the team to actually sit down and start working on. If you want a feature badly enough, you'll go build it. It turns out that this is a great filter for what is worthwhile.

Some features (like Pull Requests 2.0) spend many months in development before they're launched. PRs took about 8 months of off and on work before they were ready. Pressure from other team members will often serve as a catalyst for the implementors to finish what they started.

More strategic decisions often bubble up to the founders and we'll make a final decision on whether we need to hire additional people to make things happen faster.

Re: Ask Tom Preston-Werner, cofounder of GitHub, anything Today, Mon 18 Oct 2010.

#67
Err the Blog had about 10,000 readers when GitHub launched and Chris and PJ already had a ton of users of their popular Rails plugins.

This was a huge audience to launch to. What advice would you give to people who want to dive in to their own company without trying to build up such a huge following before hand?

Is it madness to try to bootstrap without establishing an audience beforehand?

Re: Ask Tom Preston-Werner, cofounder of GitHub, anything Today, Mon 18 Oct 2010.

#68
post #43
post #36

One thing I love about GitHub is that you embrace NIH in a good way - lots of cool technology has come out of GitHub because you were unhappy with the current solutions out there. What else have you cooked up internally that you might release to the world one day?

We have an awesome tool we call Haystack that we use to collect, categorize, and search exceptions and other notices. It uses MongoDB capped collections to keep the number of stored entries to a manageable level. "Needles" are simple JSON hashes and the views are customizable with Mustache. Exceptions can be rolled up based on a custom value provided by the producer. For instance, we roll up exceptions by filename an…

Do you think Haystack will ever be released to the public? If not, could you guys post a blog post on the GitHub blog or one of your personal blogs with more details? It sounds very interesting.

Re: Ask Tom Preston-Werner, cofounder of GitHub, anything Today, Mon 18 Oct 2010.

#69
post #63

DO you see Github as an "open web" player? What would you think if someone built tools to federate Github and Gitorious, or unfuddle? Would you help them?

I want all of the data on GitHub to be open. We recently made our wiki system Git-backed so that users could get their data. I want to do the same for Issues and comments and everything else. These things take time.

I'm not really sure what you mean by "federate Github and Gitorious, or unfuddle".

Re: Ask Tom Preston-Werner, cofounder of GitHub, anything Today, Mon 18 Oct 2010.

#70

What's the status of BERT-RPC and its usage at GitHub? Have you noticed any drawbacks to it since you first created it?

We still use BERT-RPC for all communications between the Rails app and the backend file servers to transfer information about Git repos. It's been working great but there are some improvements I'd like to build into a BERT-RPC 2.0 spec (async messaging ids for one). Overall it's been really good for what we use it for.
Post reply on HN