Fork me on GitHub
#clojure
<
2018-06-23
>
imetallica01:06:05

It worked, thanks @noisesmith! ๐Ÿ™‚

๐Ÿ‘ 4
ackerleytng03:06:40

is there a recommended way to avoid shadowing in names? in python, if you have to create a variable called list, you'd name it list_ instead to avoid shadowing python's core function

andy.fingerhut07:06:37

Many people will intentionally abbreviate names they don't want to shadow, e.g. lst

ackerleytng07:06:36

ah ok, thanks!

emil0r08:06:04

I typically put an underscore in front. So, _list

emil0r08:06:40

For python. In Clojure I put a hyphen

slipset11:06:04

One might argue that a var called list is a badly named var. If itโ€™s just holding any generic list, Iโ€™d go for l. If it ere a list of foos, Iโ€™d call it foos.

pasi12:06:49

hm, if I have a fairly large module (specifically something that generates a single pdf file) and lots of small functions that create parts of it, should I use dynamic binding to pass a reference to the data structure those functions pull their data from (it's the same for all) or should I just keep passing it down?

dominicm13:06:58

A generic list is xs

zara17:06:59

Hello guys, I was trying to setup a Solo datomic cloud environment. I was able to establish a connection through the bastion server but when a run a command I get a "Forbidden to read keyfile a". I tried to setup the IAM policy, but I am not sure where I should attach the policy to GetObject from S3.

valerauko18:06:19

it'd be great if #() anon functions worked well with the threading macros... such a shame i need to wrap them in extra parens

jrychter18:06:44

I'm getting java.lang.RuntimeException: Method code too large! errors in namespaces where there are no large methods, but lots (as in, thousands) of definitions. Is there an upper limit on the number of top-level forms?

andy.fingerhut18:06:46

@jrychter Any deeply nested doseq uses in there? Or using the core.match library or some other library that has small macro invocations that can expand into big code?

andy.fingerhut18:06:36

I would not be surprised if there was something about have many defs in the same namespace alone that causes that error, but not familiar with what it would be.

jrychter18:06:52

@andy.fingerhut no โ€” just defs. These are auto-generated definitions for FontAwesome, later used by some clever macrology. Here is an example:

(def pen-square-solid "@noinline" "M400 480H48c-26.5 0-48-21.5-48-48V80c0-26.5 21.5-48 48-48h352c26.5 0 48 21.5 48 48v352c0 26.5-21.5 48-48 48zM238.1 177.9L102.4 313.6l-6.3 57.1c-.8 7.6 5.6 14.1 13.3 13.3l57.1-6.3L302.2 242c2.3-2.3 2.3-6.1 0-8.5L246.7 178c-2.5-2.4-6.3-2.4-8.6-.1zM345 165.1L314.9 135c-9.4-9.4-24.6-9.4-33.9 0l-23.1 23.1c-2.3 2.3-2.3 6.1 0 8.5l55.5 55.5c2.3 2.3 6.1 2.3 8.5 0L345 199c9.3-9.3 9.3-24.5 0-33.9z")

jrychter19:06:18

I am not trying to determine the number of those that causes the problem.

andy.fingerhut19:06:38

binary search is your friend, in case you are curious to do so.

jrychter19:06:11

That is exactly what I'm doing right now โ€” but it still takes a while, as each compilation run takes several minutes (it's a large project).

jrychter19:06:51

I'm puzzled, though โ€” this ns contains just the ns definition: (ns ui.fontawesome.paths (:refer-clojure :exclude [key filter list clone comment import print map sort sync repeat])) โ€” and the defs. There are no functions in there.

jrychter19:06:00

And I was pretty happy with the whole setup (the resulting ClojureScript only includes code for the icons that are actually used), until I updated Font Awesome, regenerated the defs, and everything broke because the authors added new icons.

andy.fingerhut19:06:28

I just created a Clojure/Java namespace with 1 million def statements, and require'd it from the REPL. It took a while, but no error.

andy.fingerhut19:06:53

I don't have a quick way to try it with ClojureScript, which may give different results.

jrychter19:06:07

Could it matter that it's a .cljc file?

jrychter19:06:08

Also, loading it interactively (through Cider) works fine. What fails is building an uberjar with :aot :all

andy.fingerhut19:06:51

I get same error when I try 'lein uberjar' with :aot :all

andy.fingerhut19:06:58

but no error with require'ing the namespace

andy.fingerhut19:06:12

I have no idea if this is a huge plumbing change for you, but an alternative to consider would be putting all of those values in a big map with keys like :pen-square-solid. However, having a single map of that huge size in your source code rather than a data file loaded at run time might lead to the same problem, so better to test out a big map in a source file before changing all the plumbing.

andy.fingerhut19:06:38

Not sure if the issue is triggered by many top level defs in namespace, or large quantity of data.

jrychter19:06:32

No, that won't do. The whole idea here is to have individual defs, with annotations, so that macros can use a single one and only the ones used get included in the resulting ClojureScript.

jrychter19:06:00

I have the limit pinned down to somewhere between 6550 lines and 6600 lines, and there are approximately two lines per def.

andy.fingerhut19:06:45

With a namespace that just has many auto-generated defs like these: (def x0 0) (def x1 1) (def x2 2) (def x3 3)

andy.fingerhut19:06:54

It fails somewhere between 2500 and 3000 of those

jrychter19:06:24

Well that's a bit of a problem. I felt quite smug about the whole system. I don't think any other language/solution can include only the actually used icons easily.

andy.fingerhut19:06:55

Sorry, I mean between 3250 and 3500

andy.fingerhut19:06:20

and more precisely, somehwere between 3270 and 3275

andy.fingerhut19:06:35

I don't know if having large strings instead of integer values would lower that number.

andy.fingerhut19:06:58

Do you need AOT?

jrychter19:06:05

Well, AOT is there mostly to avoid surprises in deployment โ€” to have a more predictable build. It's not there for performance, which I don't really need.

jrychter19:06:32

Ok, I got the exact number. ns form + 3274 defs is the maximum I can AOT compile. Adding the next def breaks things.

jrychter19:06:23

I think I can work around this by splitting the fontawesome library by categories, but this isn't work I expected to do ๐Ÿ˜ž

dpsutton19:06:32

i'm sure bronsa could answer this but what does a namespace of def forms get compiled into? Seems strange it would hit a method too large error

jrychter19:06:18

That's why I'm wondering where to ask about this next. Perhaps it's something that can easily be fixed or worked around. I know it isn't a frequent use case, but it's a good one.

bronsa19:06:11

too many constants perharps

dpsutton19:06:58

are constants declared inside of a method though?

dpsutton19:06:25

or perhaps assigned in the constructor?

seancorfield21:06:35

I would imagine it's the static class initializer method that is triggering that -- and the def initialization would be part of that. I'm guessing but it "makes sense" to me.

alexmiller23:06:31

yes, there is I believe a ticket for this issue - itโ€™s in the init methods if I recall

alexmiller23:06:22

the naive solution to this problem is: donโ€™t do that