Could you clarify a bit the state of JS on ClojureScript land?
I just spend a weekend watching videos about Om Next and Devcards. I got Om Next inside Devcards working, but the journey from there to getting my JS based React components there was less than smooth.
I believe that things like GraphQL, Relay and Redux could provide what I have been missing on my web development story, namely code working with the data regardless of client/server separation. Om Next looks to give these things with a better language.
I'm now starting to build things with Om Next, since playing with Cljs on Devcards was really fun. I just had to rage quit a couple times trying to use my existing stuff.
I saw support for directly importing JS modules mentioned, but I could not find docs on how it could be used. Other option was to modify build configuration to get my JS files inserted into package space or converting them (I assume) to Clojure packages with CLJSJS. Finally I just mangled several build configs together which somehow worked.
I realize trying to keep one feet on JS land and other on Cljs side might not be high in your priorities, but my use case happens to require this. I keep trying, though, so thank you very much for what you've done.
I have what could be a newbie/clueless question: with Om and Clojurescript, is there a way to automatically add dependencies in parts of a project, like 'npm i -S package_name> does for package.json rather than having to add those manually in a file together with the version number?
I also have a deeper question: you were saying Falcor's reliance on strings for the query types isn't ideal and Om Next would do that differently. How? Will it be a type of binary mapping for greater efficiency?
Also, a suggestion for the cljs quick start: make it a step-by-step which works straight away for newbies (and move the deliberate trip-up with a missing inclusion to the end, perhaps in an appendix for 'troubleshooting').
Kudos for all the great work on Om, cljs and the great presentations that I usually look for on Youtube. Your demo of an old Apple with BASIC reminded me of how I started in the 80's too: Oric Atmos 48K and BASIC with immediate feedback (and line numbers, RUN and GOTO!). This made me understand why I like REPLs and the immediate feedback environments so much and I even found an Oric Atmos emulator for Mac OS X that I installed and re-coded my first program in - a white circle on a black background (and thought the web/ui thing was badly engineered until the React guys came along) :P
I use Reagent (another Clojurescript React library), and it seems this might widen the gap between the two in terms of ease-of-getting-started. Reagent is dead simple to get going with, and requires you learn basically nothing to build even moderately complicated things. Om is... somewhat notorious for having more complex abstractions built around deep thought about maintaining e.g. a single point of mutable data. This new stuff is... not something I think I could get right the first try.
That all being said, David Nolan is clearly a gamer -- Om will be interesting to watch even for people who aren't using it.
Reagent is very nice and by itself very easy to get started with.
That said, I think Om Next is more comparable to Re-frame (https://github.com/Day8/re-frame) as far as being more opinionated about handling data flow.
In that comparison, I don't think it involves any more complex abstractions or deep thought. In both cases, there is some of that, but both cases are also shining examples of explanatory documentation, which to me were edifying to read because they turn those abstractions into concrete implementations, with examples, and explain the deep thought in a clear manner.
If you use re-frame with reagent (and you should) then the ideas aren't that difficult to grasp. Om.next is more oppinionated regarding where you're allowed to run queries (subscriptions in re-frame) and transactions (handlers in re-frame), and forces you to build a single large query from the root component (wheras re-frame allows you to run several queries from every component).
As the author of Coils, another Clojure framework (coils.cc) I am really impressed to see Om Next, and I think that Om Next will become the de facto framework for building interactive websites
What about performance? How many data queries can it handle per component, realistically? How many connections can it handle? I.e., where will this solution break down?
Are the transactions (issued from within the browser) performed optimistically? If so, will this not give "flicker" in the browser in case a transaction needs to be rolled back? Is it also possible to easily perform a transaction from the server (i.e., not optimistically)?
Don't understand the question. Om.Next allows you to have a single root component, that component is responsible for building a query for every sub-component. So you really only have a single query, then render based on the results of that query. Whenever data that is touched by the query changes, affected components are re-rendered. I would also assume that the time it takes to render the view is much greater than the time it takes to run a query, although that depends on your app/query.
> Are the transactions (issued from within the browser) performed optimistically? If so, will this not give "flicker" in the browser in case a transaction needs to be rolled back? Is it also possible to easily perform a transaction from the server (i.e., not optimistically)?
I was at the Lisp NYC talk where David talked about Om Next. He specifically spent some time talking about this and demonstrated how to perform optimistic updates that do not flicker.
I can't really say anything else (the talk was great, but way above my head at this point), but it seems to be able to do that.
swannodette|10 years ago
Happy to take any questions.
jarpineh|10 years ago
I just spend a weekend watching videos about Om Next and Devcards. I got Om Next inside Devcards working, but the journey from there to getting my JS based React components there was less than smooth.
I believe that things like GraphQL, Relay and Redux could provide what I have been missing on my web development story, namely code working with the data regardless of client/server separation. Om Next looks to give these things with a better language.
I'm now starting to build things with Om Next, since playing with Cljs on Devcards was really fun. I just had to rage quit a couple times trying to use my existing stuff.
I saw support for directly importing JS modules mentioned, but I could not find docs on how it could be used. Other option was to modify build configuration to get my JS files inserted into package space or converting them (I assume) to Clojure packages with CLJSJS. Finally I just mangled several build configs together which somehow worked.
I realize trying to keep one feet on JS land and other on Cljs side might not be high in your priorities, but my use case happens to require this. I keep trying, though, so thank you very much for what you've done.
YashN|10 years ago
I also have a deeper question: you were saying Falcor's reliance on strings for the query types isn't ideal and Om Next would do that differently. How? Will it be a type of binary mapping for greater efficiency?
Also, a suggestion for the cljs quick start: make it a step-by-step which works straight away for newbies (and move the deliberate trip-up with a missing inclusion to the end, perhaps in an appendix for 'troubleshooting').
Kudos for all the great work on Om, cljs and the great presentations that I usually look for on Youtube. Your demo of an old Apple with BASIC reminded me of how I started in the 80's too: Oric Atmos 48K and BASIC with immediate feedback (and line numbers, RUN and GOTO!). This made me understand why I like REPLs and the immediate feedback environments so much and I even found an Oric Atmos emulator for Mac OS X that I installed and re-coded my first program in - a white circle on a black background (and thought the web/ui thing was badly engineered until the React guys came along) :P
Fun years ahead.
hawkice|10 years ago
That all being said, David Nolan is clearly a gamer -- Om will be interesting to watch even for people who aren't using it.
jraines|10 years ago
That said, I think Om Next is more comparable to Re-frame (https://github.com/Day8/re-frame) as far as being more opinionated about handling data flow.
In that comparison, I don't think it involves any more complex abstractions or deep thought. In both cases, there is some of that, but both cases are also shining examples of explanatory documentation, which to me were edifying to read because they turn those abstractions into concrete implementations, with examples, and explain the deep thought in a clear manner.
Skinney|10 years ago
ams6110|10 years ago
amelius|10 years ago
Gonzih|10 years ago
intellectable|10 years ago
[0] https://chrome.google.com/webstore/detail/react-developer-to...
zubairq|10 years ago
amelius|10 years ago
Are the transactions (issued from within the browser) performed optimistically? If so, will this not give "flicker" in the browser in case a transaction needs to be rolled back? Is it also possible to easily perform a transaction from the server (i.e., not optimistically)?
Skinney|10 years ago
sotojuan|10 years ago
I was at the Lisp NYC talk where David talked about Om Next. He specifically spent some time talking about this and demonstrated how to perform optimistic updates that do not flicker.
I can't really say anything else (the talk was great, but way above my head at this point), but it seems to be able to do that.