A luxuriously simple and powerful way to make front-ends with DataScript and Reagent in Clojure.
@timgilbert Direclty using Datomic filters could be bad if you tried to reexecute them on every transaction. However, you don't necessarily have to. If a filter function is pure, it takes a datom and decides whether or not to keep it. That's it. So when you listen for transactions you should be able to just take the new datoms and apply the filter function to them. Simple. The problem is when the filter function needs to change over time. Which it should in any case where permissions for a datom can change over time (say, you grant someone access to datoms you created). For this, I've been thinking about how we can build more expressive filter views based on datalog queries, where changes to the queries trigger recomputations on the filters.
That having been said, I would love to hear how @seantempesta has been using filters, and how it's worked out.
a question though
my database is a bunch of entities which represent instructions for an engine on the backend to do some calculation based on the instructions
so for a given class of instructions I get all the eids with an attribute which all those instructions have
with posh/q
then I do a bunch of posh/pulls on the eids in the parent component and pass the entities into some component which renders the instructions in some way
so this turns out to be a bit slow because there's a pull for each eid
the problem is, that the initial posh/q which gets all the eids will rerun whenever I edit any of the entities
when in reality it would only need to rerun when an entity is added or retracted
so in essence
I have a posh/q which gets all the eids I want
I then get the data for those eids with a bunch of posh/pulls
even though it's going to get the same list of eids because no entity was added or retracted
@UrKr There was some talk about bringing back manual pattern matching, which would let you specify that the q should only be rerun if a transaction looked like it was going to create a new entity that you'd want to match on. I don't know if that ever happened, but it would be worth looking into.
@metasoarous I mean, to be able to be used in a bootstrapped cljs environment like lumo or in my use case a browser based generative UI exploration tool using klipse/replumb + reagent + posh + datascript + spec . Day8/re-frame#325 is a pull request that achieves that for re-frame. Live coding UI in the browser is the ultimate addiction :)
I actually forked posh and started looking into this yesterday... It seems like the clj macros in base_plugin....What is the setup for development of the reagent parts? Reagent is not in the dependecy list... sorry ..maybe i am doing something wrong ...
I actually forked posh and started looking into this yesterday... It seems like the clj macros in base_plugin....What is the setup for development of the reagent parts? Reagent is not in the dependecy list... sorry ..maybe i am doing something wrong ...
True dat :-)
Reagent is not in the dependency list so that in theory it could be used with another front end framwork
It's been abstracted that way
So you have to require reagent separately if you want to use the reagent namespace
@metasoarous My slow brain will certainly need some guidance. I haven't traveled much in macro land!
What branch should i work off ? I am getting a bunch of errors when compiling from the master branch? this one is the first that pops up, it seems aget is missing an argument.
What branch should i work off ? I am getting a bunch of errors when compiling from the master branch? this one is the first that pops up, it seems aget is missing an argument.
@metasoarous thanks!
Hi @alexandergunnarson, I am getting a few errors when compiling from master first one is this one , could you give me some orientation on which branch is best to work off? thanks!
Hi @alexandergunnarson, I am getting a few errors when compiling from master first one is this one , could you give me some orientation on which branch is best to work off? thanks!
@metasoarous cljs.core/aget inside aget is being called with 1 argument, not aget itself. the java .get method is getting both arguments passed to aget*
I can't look at that right now, but that will help once I or someone else does have the time to
Keep us posted!
that aget call missing an argument was breaking compilation. I just fixed that and compiles ok with some warnings that i am looking into now.
they seem to originate from the
Use of undeclared Var posh.lib.ratom/array at line 202 src/posh/lib/ratom.cljc is an example.they seem to originate from the
if-cljs macro. investigating now, need some coffee to un-bend mi mind :)
Heh; macro recovery