Hacker News new | past | comments | ask | show | jobs | submit login
SQLite with Richard Hipp [audio] (changelog.com)
109 points by grey-area on May 14, 2016 | hide | past | favorite | 16 comments



"The reason why ... [SQLite] needed to exist ... The genesis story?" Richard summarizes the development history. 20:39 to 27:20

"Why is SQLite so ubiquitous?" "Its goal is that it just works .... Apparently we win competitions ... I don't know why or how." 30:20

"SQLite is ... not client-server. Why didn't I do this? ... I'm not a database person. I didn't know I was supposed to." 34:25

On "flexible typing" 40:22

"People tell me I could have made a lot of money off this if I had any business sense .... But you know what? We make enough." at 1:06:10


Thanks for taking the time to share your favorite quotes.


I am curious about the third one.

Why do all other DB products use a clinet-server model, or at least have a dedicated thread for the DB?

I suppose, you can handle writes asynchronously to gain some performance. Are there other upsides?


He gives the answer himself a few minutes later (46:00):

"You need that server process to coordinate write concurrency."


Thanks for pulling out all these time codes and quotes. This was awesome.


I found the final part of the podcast [1] where DRH talks about having a design for a "Git killer" intriguing. He describes it as a possible future project distinct from Fossil, which he calls only an incremental improvement upon Git. I would have liked to hear more of a hint at the novel design involved, though. Before that DRH discusses his early career and how SQLite and Hwaci came to be, which was also interesting.

Overall, I thought the podcast was well worth listening even if you're reasonably familiar with SQLite itself and its history.

[1] Approximately from the 01:17:00 mark on.


Has someone a speech regognition software like Dragon NaturallySpeaking on hand to create a transcript? 1:23 is too long, and a transcript would be handy to skip through the interview.


I tried CMU's Sphinx without training it first.

  $ ffmpeg -i changelog-201.mp3 -acodec pcm_s16le -ac 1 -ar 16000 out.wav
  $ pocketsphinx_continuous -infile /tmp/out.wav > /tmp/transcript.txt
The first line of output read

    i'm richard have been you're listening to the change log and have it type collector who won this bit she's long time they're both have people get this this episode two zero and the invaders now i've got to purchase this creature
If anyone knows how to train Sphinx to understand technospeak, I would love to know also.


I found this episode very interesting, especially the discussion of the long term support contract required by Airbus since SQLite is used on some aircraft.


In which minute is that?


1:10:20


If anyone's after a decent cross platform GUI for SQLite, this one's good:

  https://github.com/sqlitebrowser/sqlitebrowser
Note - Self promotion, as I'm a committer on that project. :)


I also listened to this episode recently. Two questions I wish they'd asked:

- What are drh's thoughts on the WebSQL kerfuffle?

- What is the future of sqlite4?


Maybe we can get Richard back on the line for a few more questions and an extended version? I'm sure he'd be down for that.


Lot's of good bits around 45:00:

"We don't compete against Oracle, we compete against fopen(3)"


Hopefully, this one is more informative, than another one from couple years ago, which was disappointingly untechnical (and SQLITE is not a trivial piece of software, despite what one may think, just try to understand the semantics of error handling when it comes to rollbacks, for example).




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: