Re:Apply HN: Brodlist – a new type of database
by david927 on Thursday April 07, 2016 @01:56AM
(#11445036)
1) C++
2) It's clearer and allows for more powerful constructs (even if they're not being talked about yet).
3) Yes.
4) It depends on the number of writes but you can adjust the level of history to the level you want.
5) Great question. I'll have to circle back to this later.
6) The feature list above beats SQL Sever on each point.
7) Initially as a service but that could change.
8) The research stems from the Kayia programming language (kayia.org).
9) The feature set for the price. It solves a lot of back-end needs and yet is much cheaper.
Re:Apply HN: Brodlist – a new type of database
by asimuvPR on Thursday April 07, 2016 @09:12AM
(#11446934)
Do you have any documentation regarding the DSL? I'm interested in looking into this. Feel free to email me (check profile) with more information.
Re:Apply HN: Brodlist – a new type of database
by david927 on Thursday April 07, 2016 @12:51PM
(#11448713)
I apologize for the lack of information on the site. I'll add a lot more to the site soon including a REPL. You can find something for now at kayia.org.
Thank you so much for your interest!
Re:Apply HN: Brodlist – a new type of database
by asimuvPR on Thursday April 07, 2016 @01:57PM
(#11449200)
I will look into it, thank you.
Re:Apply HN: Brodlist – a new type of database
by smt88 on Thursday April 07, 2016 @02:05AM
(#11445076)
Why didn't you use Rust? (I swear I'm not a troll, and I'm genuinely curious.)
Re:Apply HN: Brodlist – a new type of database
by sumodirjo on Thursday April 07, 2016 @06:13PM
(#11451118)
Why rust? Why not using Go instead of C++
Re:Apply HN: Brodlist – a new type of database
by smt88 on Thursday April 07, 2016 @06:42PM
(#11451285)
Why Rust? Because Rust has excellent concurrency support and memory safety, both of which are useful for databases.
For something as low-level and performance-intensive as a database intended to replace MySQL/Postgres, Go wouldn't work because it's garbage collected.
Re:Apply HN: Brodlist – a new type of database
by asimuvPR on Thursday April 07, 2016 @07:56PM
(#11451636)
I feel Rust is still not old enough to be used on databases. It is a relatively new language with an amazing future. But it might not be a good choice for a new product that is meant to be the workhorse of a system. C++ is the correct and safe choice under the circumstances. Good question though. Its always beneficial to question architectural choices in order to see the other side of the coin. :)
Re:Apply HN: Brodlist – a new type of database
by smt88 on Thursday April 07, 2016 @08:08PM
(#11451696)
It might be because I was watching MemSQL for a long time, but I think Brodlist is probably years away from being used in production. Rust is stable now, so I think there's an acceptable tradeoff between the risk of Rust lacking an ecosystem and the benefits of using Rust.
Anyway, it doesn't matter that much in the end. I was just curious.
2) What are the key benefits of designing a new DSL for this database?
3) Given that it uses a built-in API, will security audits results be made public?
4) It seems to save "snapshots" of different moments in time. How does that affect database size? Does each snapshot act as its own database instance or is it just a data backup?
5) How do you plan to introduce and train programmers into using various different concepts at once? What kind of strategy do you plan to leverage for mind share?
6) You mention replacing the relation database. Explain how this is better than Microsoft's SQL Server. [Comparing a paid product against another rather than to a free one.]
7) Will this be offered as a stand alone product or as a service?
8) What other databases (or systems with similar complexities) has the team worked on / built ?
9) What makes this a must have standout product from other databases out there?