BitShares Forum
Main => General Discussion => Topic started by: hadrian on March 05, 2015, 05:45:51 pm
-
I saw this (https://www.maidsafe.org/t/rust-vs-c/3216) topic in the maidsafe forum. I think it's interesting enough to be a topic here in "General Discussion".
Essentially, it seems from David Irvine's comments that there's a possibility of their platform switching from C++ to Rust! Can you imagine what a kerfuffle it'd cause now if our very own bytemaster made similar comments on our forum?!
David Irvine is the main man at maidsafe, and he is 'dirvine' on the forum... have a read!
I've created this post, because I expect people here will see at least a couple of discussion points:- The fact that David Irvine has made such comments (it's relevant to where we were and where we're going)
- Technical aspects - is Rust really THAT good? Is this something we should or could consider at any point?
I'm actually pretty interested in the second point, especially as I think the first has already been covered to an extent elsewhere.
Although I'm virtually clueless about programming languages (I've just started to learn Python), I'd like to hear from people in the know.
Also, feel free to tell me to learn something other than Python. :P
-
Rust is really that good.
Nevertheless maid took a big dive after the post, not unlike what happened to bts.
-
Nevertheless maid took a big dive after the post, not unlike what happened to bts.
Doing what some expect the market might want, is one of the great fallacies of our time. Knee-jerk pop-politics and businesses looking simply to maximise stocks value, creates a world of trouble. Better to be authentic and genuine, if you are pitching at long term success.. let the market follow.
-
- Technical aspects - is Rust really THAT good? Is this something we should or could consider at any point?
I've been following Rust for a while with extreme interest and would absolutely consider it as my #1 choice for a FUTURE (after a year or two) project. It still has not hit 1.0 and I would expect another year or two for an an acceptable ecosystem of libraries, etc. to develop.
-
Rust is pretty nice... but not mature enough yet in my opinion.
-
- Technical aspects - is Rust really THAT good? Is this something we should or could consider at any point?
I've been following Rust for a while with extreme interest and would absolutely consider it as my #1 choice for a FUTURE (after a year or two) project. It still has not hit 1.0 and I would expect another year or two for an an acceptable ecosystem of libraries, etc. to develop.
+5% My hope is that in a couple years we take all the lessons learned from BitShares development and usage and reimplement the entire platform the best way we know how at the time, this time written entirely in Rust, and release it as BitShares 3.0 whose genesis block contains a commitment to a genesis database state which is a live snapshot of the most recent state from the BitShares 2.x chain it instantaneously replaces.
-
- Technical aspects - is Rust really THAT good? Is this something we should or could consider at any point?
I've been following Rust for a while with extreme interest and would absolutely consider it as my #1 choice for a FUTURE (after a year or two) project. It still has not hit 1.0 and I would expect another year or two for an an acceptable ecosystem of libraries, etc. to develop.
+5% My hope is that in a couple years we take all the lessons learned from BitShares development and usage and reimplement the entire platform the best way we know how at the time, this time written entirely in Rust, and release it as BitShares 3.0 whose genesis block contains a commitment to a genesis database state which is a live snapshot of the most recent state from the BitShares 2.x chain it instantaneously replaces.
My hunch is major and sudden changes like that won't be possible if ecosystem businesses are large enough - another way, how much effort would it take all Bitcoin businesses to upgrade to an entirely newly designed blockchain
Sent from my iPhone using Tapatalk
-
My hunch is major and sudden changes like that won't be possible if ecosystem businesses are large enough - another way, how much effort would it take all Bitcoin businesses to upgrade to an entirely newly designed blockchain
Downloading and installing a new piece of software? Assuming the RPC API was unchanged and the forking was instantaneous, I can't imagine it would be much effort for users to upgrade their clients. Also, the upgrade would only go into effect when more than 75% of BTS stake votes in favor of the upgrade.
We need to remain agile enough to make large, significant changes gradually over time with smooth transitions. Delegates and stake voting help with this process. We don't want the technology to ossify and stagnate like Bitcoin because then that leaves us open to inevitably being replaced by a better technology despite any advantages in network effect.
-
My hunch is major and sudden changes like that won't be possible if ecosystem businesses are large enough - another way, how much effort would it take all Bitcoin businesses to upgrade to an entirely newly designed blockchain
Downloading and installing a new piece of software? Assuming the RPC API was unchanged and the forking was instantaneous, I can't imagine it would be much effort for users to upgrade their clients. Also, the upgrade would only go into effect when more than 75% of BTS stake votes in favor of the upgrade.
It would be much effort to deal with the 1000's of bugs this kind of transition would create, as the difference in language features expose/hides edge cases. IMO this idea is completely insane unless the project is in its infancy.
-
Rust is pretty nice... but not mature enough yet in my opinion.
Yes, Rust is in its beta and there are likely to be changes in it. The market fears that the maid's move to Rust would cause further delay to the project (another 1-2 years??). The fear translates to a big sell-off.
-
Or is a port to Rust just an excuse to buy some more time?
-
Or is a port to Rust just an excuse to buy some more time?
:D
-
BTW, Codius is also moving to Rust: https://codius.org/blog/codius-rust/
-
I played a bit with Rust and really like the major feature of using linear types to manage lifetimes.
But I wonder, given how important the web is, whether using a language that can target both native code and javascript would be better. The web is still very immature with xss attacks, poor RNG implementations, Javascript's primitive feature list - but it's what end-users want. With websockets it's probably even possible to proxy the p2p.
Reimplementing for each platform like bitshares has done with the main repo and then with bitshares_js is a lot of duplication and prone to error.
-
Rust is pretty nice... but not mature enough yet in my opinion.
What is your opinion of Flow Based Programming mods and approaches such as Flow? https://foundationdb.com/key-value-store/white-papers/flow
Looking at the core Bitshares core I always think "That should have been FBP".