02:43 <endformationage> SeanTAllen: Will, do.

03:48 <SeanTAllen> thanks endformationage

04:06 <endformationage> SeanTAllen: https://github.com/ponylang/ponyc/issues/1994

12:16 <SeanTAllen> thanks

18:07 <JordanUK> hi guys does anyone know how to return a 'this' in a function but of a ref capability? i know 'this' always returns a tag capability

18:08 <JordanUK> for example i have a tree of nodes with - fun ref parent(): Node ref , that returns its parent node. But for the root node of the tree, I want the function to return 'itself' as a ref

18:09 <jemc> JordanUK: I think I'm missing context in your question - the statement that `this` aways returns a `tag` capability isn't strictly true

18:09 <jemc> could you give a full example of the code you're compiling, and the error you see?

18:16 <JordanUK> so in my Node object i have a field _parent: Node ref; and i have a specific constructor for a Node object that is the root, in that constructor I tried to do _parent = this; but the compiler complains Node tag is not a subtype of Node Ref

18:20 <JordanUK> https://pastebin.com/PUWKj6YR

18:59 <SeanTAllen> ah so JordanUK

18:59 <SeanTAllen> the problem is that until an object is fully initialized, "this" is a tag.

18:59 <SeanTAllen> you have to initialize all the fields first in your constructor

19:01 <SeanTAllen> So JordanUK, when you call parent() on the root node, do you really want it returning itself?

19:02 <SeanTAllen> It seems like you want `let _parent: (Node ref | None) = None` or something like that for the typ

19:13 <jemc> JordanUK: in this case, the issue is that you're in a constructor, where not all fields have been assigned yet - in such cases, `this` will be `tag` - the error message is supposed to tell you that it would be possible if all fields were assigned by that point

19:15 <SeanTAllen> i think we have a ticket open for that change in error message, yes jemc?

19:17 <jemc> SeanTAllen: https://github.com/ponylang/ponyc/issues/1866

19:17 <jemc> it's actually a regression, I think - that clarification used to exist

19:17 <jemc> not sure how/why it disappeared

19:19 <SeanTAllen> hmmm... i wonder how we would prevent regressions like that

19:23 <jemc> we currently test error messages (for new tests - many od tests do not test error messages), but we don't currently have a way to test the "error continuations" that provide extra info for an error

19:24 <jemc> in ponycc I'm testing the entire error output, but in ponyc we currently do not

19:37 <JordanUK> ic thanks for all the help and yes i worked around this with a None type and i also understand why it has to be of a tag cap until the object is fully constructed

19:39 <jemc> I'll try to make time this weekend to fix that error message, since it seems like a common pain point

22:47 <cquinn> question about the files package: If I have a FilePath representing a file, what is the best way to get a FilePath to its containing directory? It seems that a FilePath.dir() or FilePath.parent() would be useful.

22:48 <cquinn> Or is their an auth/caps problem walking up the hierarchy that way?

22:51 <SeanTAllen> FilePath is designed to keep you constrained to that Path and below.

22:51 <SeanTAllen> so yes to the auth/caps issue cquinn

22:51 <cquinn> ok, cool. It just came to me as I was assembling the question :)

22:52 <cquinn> I will keep the dir as a FilePath and the file as a String below it. Or similar.

22:56 <cquinn> oh, hey, while I'm thinking about it. Is there a way to run ponyc without the optimizations and without debug generation? Just for quick iteration.

22:57 <SeanTAllen> you mean, just to see if it compiles?

22:57 <cquinn> yes, compile & run for testing out things

22:57 <cquinn> the optimizer takes a long time, and I'm not usually debugging

22:58 <SeanTAllen> i use pony --debug for that. not sure if what the step would be

22:59 <cquinn> I do the same, ponyc -d, but I was just wondering if there was a way to get the fast compile without the symbols out. might be slightly faster too.