Clojurians Log v2

I’m on master branch

but I keep getting that error with the Id from server console

strange

I’m working with another person and he has the same issue

maybe look at the network request payload to see what the server is sending the client?

make sure you’ve cleared any browser cache

not that this sounds like that kind of problem

develop branch does look broken in general

with advanced opt...prob a new client prob with optimizations

works for me on develop with dev build

but not production

Ok, I’ll switch to develop and try that branch

make sure you clean, and make sure you clean browser cache.

using chrome?

yes

open console. click options

find "disable cache while js console open"

turn that on, and leave console open

reload

master should be working better than develop

I gotta head out though. But I have verified that develop dev build worked for me. I build the support and dev build, then loaded it from the server port (which the server says when it starts). Nothing works right if you don't do that, though 😉

Ok great, thank you very much for your prompt reply! I’ll try all that

@mduthey has joined the channel

@jonystorm did you get it working?

Nope, I thought it might be datomic edition, since I have pro, but I keep getting the same error message

did you modify something? It should be using datomic free

I added

:repositories {"" {:url ""
                                   :creds :gpg}}

try running it stock. shouldn't matter, but it does work for me.

I’ve just tried with develop branch and it’s the same result

hm. That's really odd.

so, your usage sequence is: 1. Start the regular app 2. Play with it...you see server accepting remote changes 3. Submit a support request via Help with a message 4. Copy the ID from the server log 5. Start the support viewer with that ID right? No pauses or restarts in the middle anywhere

and you're building the support and dev targets via figwheel or something

I see stuff like this in my app-state:

[:sale.issue/by-id 17592186045605] {:ui/fetch-state

1. I run JVM_OPTS="-Ddev -Dtest -Dsupport" lein run -m clojure.main script/figwheel.clj` 2. once it finishes I run (on another console) lein run and when ready (go) 3. go to play with it, all changes ok 4. submit support request via help with message 5. copy the id from server log 6. open new browser tab on and get that error, no restarts no pauses

@mitchelkuijpers try snapshot and let me know

Sure

@tony.kay totally fixes it awesome

@jonystorm Did you actually play with the app between 3 & 4

?

If not, there are no steps saved...the iseq message is basically saying there is nothing to play back

just did your exact steps and it works

I added a couple of elements to the list and marked them as done

At the figwheel prompt, maybe try (reset-autobuild) and make sure you've got a good build?

Also, try a different browser. Could be a chrome plugin hosing you?

I've seen ppl get screwed by certain dev cljs plugins

I’ll try disabling all chrome plugins and also will try on a different browser

Also, if you can show me (via the network tab) what is coming over the network when you try the support viewer, it would be potentially enlightening

e.g. what is the response...it is transit+json, so it isn't pretty

The snapshot seems to break something else related to loading markers

Hm. bummer. let me know what you're seeing. The merge for ident-based remotes was wrong (and still has a problem). Could be related. If you can quantify that more, it would help

Sure I am going home now but I will check it out tommorow

It seems something todo with lazily-loaded

k

But the other bug is fixed ^^

I tried firefox and safari, disabling chrome plugins and it’s the same result.. regarding the response, on the support view I’m not getting any transit+json response, no requests to the api, the response to the support.html request is the html that loads the support.js

you should see an /api request

I cannot even guess what is going wrong for you. If you're not seeing an API request, then it means it isn't even figuring out the URI...that makes no sense

my guess is that the error breaks execution so it doesn’t even try to send those api requests

I'd like to see the tail of your network requests for support

no search

the requests in between are all js files

Internal redirect???

you have a proxy on your network?

or some other network magic going on?

you're getting a 307 from your localhost, which almost always means you've got some screwy OSX thing installed that is messing with localhost connections

the 307 is not coming from your untangled server

but I’ve just checked on firefox (where I’m getting the same error) and shows no internal redirect

no proxies, nor network magic

Testing on 2 other pc’s.. both same error, one of them has the 307 redirect, the other doesn’t. We will keep testing and share if we get it to run correctly. Thanks for the concern

correct server port?

strange

yes

PCs, not mac?

3 macs

ok

the 307 should not be there...um, figwheel does try to connect on the generated js. I assume figwheel is still running

right.. it is, no server errors, nor repl nor figwheel

is there a way to check the databse on ram? to see if it is actually saved ?

I imagine that if I get the “new support request id” it is because it’s returned by datomic and it is saved..

the problem isn't with your server if you never ask the server

according to your js console, you never get to that pt

so debugging datomic is a waste of time at the moment

right.. so the error is because of the missing api request, there’s something preventing it, and it’s not the error

could you use quicktime to screen record, doing the following: 1. Do everything up to the support.html?id=1001 bit 2. Make sure network dev tool is open 3. scroll slowly through the list of resources loaded (so I can pause and look)

then maybe use handbrake to drop the video to mp4 and 5fps (so it is small) and send it to me?

or, we could spin up hangouts and look via screen sharing

I have no more time to fiddle via text messages 🙂

@dpineda has joined the channel

looks like a bug in 0.8.1-SNAPSHOT (load markers are messed up). Using an older version of client fixes it (todo mvc)

Something goes wrong with the replacement of tempid's with follow-on reads after mutations

which version?

0.8.1-SNAPSHOT

thanks

And I seem to have to add a extra key to refresh 0.8.1 to make sure a certain key is re-read after a load finishes

But fixed it by adding a extra refresh key (so maybe a user erro)

s/erro/error

Ow lol this is not hipchat lol

the latter might actually be due to us removing the forced root re-render

Ah but it worked on 0.8.0

I thought that was removed on 0.7.0

which I thought we had already affected

yeah

don't know about that one

we didn't change more about refresh

since that

I'll see if I run into that some more

thanks

I am looking into where the follow on read tempid-replacement might go wrong

Hmm I am doing a mutation and then a separate df/load in one function maybe that is problem

it's ok to have multiple mutations on a single thread execution

but the mutations will be ordered before the loads

in the net queue

(assuming they are remote)

Yeah that is what I thought

Yes they are

This seems correct:

[2609.067s] [om.next] transacted '[(sale/link-issue {:issue-id "10103", :sale-id 17592186045629, :issue-link-id <#C06DT2YSY|om>/id["3fe6d980-7792-40a6-a6ea-796a13608702"], :funnel-id 17592186045568}) [:sale/by-id 17592186045629]], #uuid "a6f5904c-ccf2-40c0-a4f6-6019788c591d"
console.js:203  [2609.073s] [om.next] transacted '[(untangled/load {:remote :remote, :refresh [], :parallel false, :fallback nil, :marker true, :post-mutation nil, :without #{}, :post-mutation-params nil, :target nil, :query [{[:sale.issue/by-id <#C06DT2YSY|om>/id["3fe6d980-7792-40a6-a6ea-796a13608702"]] [:db/id :ui/fetch-state :sale.issue/issue-id {:sale.issue/funnel [:db/id]} :sale.issue/issue [:key {:fields [:issuetype :project :assignee :status :summary :priority]}]]}]}) :ui/loading-data], #uuid "b6ff70e3-b94d-416f-b5d3-a20bc812535f"

["^ ","~$sale/link-issue",["^ ","~:tempids",["~#cmap",[["~#om/id","~u3fe6d980-7792-40a6-a6ea-796a13608702"],17592186045645]]]]

weird, but it worked in 0.8.0?

cause that would be the networking upgrade not remapping over the network queue, and that was an earlier release I thought

Yes it works in 0.8.0

I'll try it once more just to be sure

one of the patches was to add support for more custom options when creating client. I suspect something is broken there, and some plumbing isn't in place

Yeah the sequential option might be related

Almost restarted the app so I can be sure

Yes it works on 0.8.0

frustrating...I cannot reproduce the errors in todomvc no matter what I try, but they work on older version as well.

I wonder if I've not released a SNAPSHOT to clojars and it is using my local update

hmmmm

@jonystorm Any chance you could re-try with 0.8.1-SNAPSHOT. I suspect perhaps I didn't push an update to clojars? Same with @mitchelkuijpers

I just pushed an update to be sure

untangled-client-0.8.1-20170404.183406-8.jar

Sure I'll try

Sure.. I’ll try right now

@tony.kay Still broken here

do you customize migrate at all?

:reconciler-options {:migrate nil
                               :pathopt false
                               :shared {;; Take absolute positioned elements in account for resizing
                                        :resizing/abs-pos-elems resizing/absolute-positioned-els-on-screen}}

migrate is nil

So no

k

@tony.kay genius!!! it works, I’m on develop branch with 0.8.1-snapshot

with the latest jar

I owe you a couple of beers at least

😕 I didn't do anything except re-push it

to clojars

still a good reason to drink 🍺

seems that 0.8.1-SNAPSHOT still has some issues, though

explains why I could not make it fail at least

@mitchelkuijpers I'll dig into the other bits...the parts that seem broken are: 1. tempid rewrite on net queue (or possibly ordering?) 2. some kind of load marker thing

The load marker thing seems to work (for my case)

The ordering seems correct because the calls go in the right order

The tempid rewrite seems to be my only issue

oh, so only the tempid rewrite?

ok, thanks

@piotrek has joined the channel

Hello! I am just getting started with untangled 🙂 Is https://github.com/untangled-web/untangled-template meant to be cloned to get started or is it somehow available as a lein template?

clone

we have a few too many resources to maintain, and lein templates are kind of pain with little gain

@piotrek

Thank you - it wasn’t clear to me

I also noticed there is bin/rename-project.sh. Is it intended to be run just after cloning?

yes, though it required a weird version of sed...I think that has been fixed. Try it..worst case you have to use git to undo 🙂

@mitchelkuijpers just out of curiosity: Do you have a custom network handler?

We added an extra protocol, that if not implemented, caused network requests to go in parallel. I'm fixing the default back to sequential.

I suspect that was the problem

fixed and pushed to clojars as a new snapshot. see if that helps

@gardnervickers @jasonjckn I think one of you were asking about load overwriting UI attributes. Found that problem and fixed in 0.8.1-SNAPSHOT

@tony.kay Oh thanks so much!