4 ms·

FriendFeed Changelog: see what code we are writing

31 points by paul 19y ago

pchristensen 19y ago
"Paul recently suggested that we could save ourselves a lot of time writing blog posts if we just sent out our changelists from Mercurial" Wow. I mean, wow!! This is a great way to get technical people invested in the development of your app!
paul 19y ago
Here's the actual changelog, btw: http://changelog.friendfeed.com/ http://changelog.friendfeed.com/
nickb 19y ago
Very cool! PS: paul, what do you guys use on the backend? I see you guys use nginx (great choice). What else?
- paul 19y ago
  Python, mostly. There's a little bit of c++ to make some things fast, but we're trying to minimize that.
  - rsa 19y ago
    Paul, Do you guys use any web framework for python ?
    - paul 19y ago
      It's custom, but apparently similar to a few of the popular ones.
michael_nielsen 19y ago
I'd love to set an ethographer loose on this, and see what insights they have about how programmers and programming works. Might make a nice PhD thesis for someone :-)
- neilc 19y ago
  There have already been a bunch of academics studying the version control histories of major open source projects, for example as a means to gain insight into why open source development "works" (i.e. is able to produce very high-quality software). One of the cool things about studying an OSS project is that everything is public, and the vast majority of the communication happens via an archived, digital medium: you can correlate VCS history with mailing list activity with bug tracker activity with website traffic logs, for example. An example paper related to this idea is http://opensource.mit.edu/papers/lakhanivonhippelusersupport.pdf http://opensource.mit.edu/papers/lakhanivonhippelusersupport... ("How Open Source Software Works: "Free" User-to-User Assistance", studying Apache)
joshwa 19y ago
I've considered doing this for a while, too. I went to far as to make a post-commit hook script for subversion to post to twitter, but it was friends-only so only my collaborator and I could see it. Even better would be to post the diffs :-P Would people be interested in seeing the svn changelog on my current project?
joshwa 19y ago
choice exerpts: changeset: 50bb94a18270 user: bret date: 2008-02-26 13:23:50 description: Undo shitty code. --- Sat, Feb 23: Let's redo our entire UI two days before launch :)
henning 19y ago
damn, that's a lot of commits. I only usually make 3-5 a day, but then again I only work on my work stuff 8 hours a day, too.
- mrtron 19y ago
  I find it depends on the type of work I am doing. Creating a new module - a daily commit. Bug fixing - a zillion commits. GUI work - a zillion commits. Since it was right before prelaunch, and presumably a lot of GUI/bug fixes, seems right.
eusman 19y ago
modern cold war
staunch 19y ago
When I think about doing this myself I immediately start coming up with flimsy excuses for why I can't. I believe a lot of doing a startup well is daring to put yourself out there to be judged. This is right in line with that idea (to an extreme). Brave and commendable.

31 points by paul 19y ago

"Paul recently suggested that we could save ourselves a lot of time writing blog posts if we just sent out our changelists from Mercurial" Wow. I mean, wow!! This is a great way to get technical people invested in the development of your app!

paul 19y ago

Here's the actual changelog, btw: http://changelog.friendfeed.com/ http://changelog.friendfeed.com/

nickb 19y ago

Very cool! PS: paul, what do you guys use on the backend? I see you guys use nginx (great choice). What else?

paul 19y ago
Python, mostly. There's a little bit of c++ to make some things fast, but we're trying to minimize that.
- rsa 19y ago
  Paul, Do you guys use any web framework for python ?
  - paul 19y ago
    It's custom, but apparently similar to a few of the popular ones.

michael_nielsen 19y ago

I'd love to set an ethographer loose on this, and see what insights they have about how programmers and programming works. Might make a nice PhD thesis for someone :-)

neilc 19y ago
There have already been a bunch of academics studying the version control histories of major open source projects, for example as a means to gain insight into why open source development "works" (i.e. is able to produce very high-quality software). One of the cool things about studying an OSS project is that everything is public, and the vast majority of the communication happens via an archived, digital medium: you can correlate VCS history with mailing list activity with bug tracker activity with website traffic logs, for example. An example paper related to this idea is http://opensource.mit.edu/papers/lakhanivonhippelusersupport.pdf http://opensource.mit.edu/papers/lakhanivonhippelusersupport... ("How Open Source Software Works: "Free" User-to-User Assistance", studying Apache)

joshwa 19y ago

I've considered doing this for a while, too. I went to far as to make a post-commit hook script for subversion to post to twitter, but it was friends-only so only my collaborator and I could see it. Even better would be to post the diffs :-P Would people be interested in seeing the svn changelog on my current project?

joshwa 19y ago

choice exerpts: changeset: 50bb94a18270 user: bret date: 2008-02-26 13:23:50 description: Undo shitty code. --- Sat, Feb 23: Let's redo our entire UI two days before launch :)

henning 19y ago

damn, that's a lot of commits. I only usually make 3-5 a day, but then again I only work on my work stuff 8 hours a day, too.

mrtron 19y ago
I find it depends on the type of work I am doing. Creating a new module - a daily commit. Bug fixing - a zillion commits. GUI work - a zillion commits. Since it was right before prelaunch, and presumably a lot of GUI/bug fixes, seems right.

eusman 19y ago

modern cold war

staunch 19y ago

When I think about doing this myself I immediately start coming up with flimsy excuses for why I can't. I believe a lot of doing a startup well is daring to put yourself out there to be judged. This is right in line with that idea (to an extreme). Brave and commendable.