Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 11:11
    dotnet-maestro[bot] commented #38172
  • 11:06
    dotnet-maestro[bot] commented #38172
  • 11:01
    dotnet-maestro[bot] commented #38172
  • 10:56
    dotnet-maestro[bot] commented #38172
  • 10:51
    dotnet-maestro[bot] commented #38172
  • 10:46
    dotnet-maestro[bot] commented #38172
  • 10:41
    dotnet-maestro[bot] commented #38172
  • 10:36
    dotnet-maestro[bot] commented #38172
  • 10:30
    dotnet-maestro[bot] commented #38172
  • 10:25
    dotnet-maestro[bot] commented #38172
  • 10:20
    dotnet-maestro[bot] commented #38172
  • 10:15
    dotnet-maestro[bot] commented #38172
  • 10:12
    jcouv synchronize #37596
  • 10:10
    dotnet-maestro[bot] commented #38172
  • 10:05
    dotnet-maestro[bot] commented #38172
  • 10:00
    dotnet-maestro[bot] commented #38172
  • 09:55
    dotnet-maestro[bot] commented #38172
  • 09:50
    dotnet-maestro[bot] commented #38172
  • 09:45
    dotnet-maestro[bot] commented #38172
  • 09:40
    dotnet-maestro[bot] commented #38172
Joseph Musser
@jnm2
Correlating other things like Task.Status with the nullability of Task.Exception
Yair Halberstadt
@YairHalberstadt
Biggest hurdles I think remain: tracking dependant variables, tracking arrays, enforcing user initialised variables, understanding generic code, understanding linq.
Solving all of those will probably take as long as the initial release took.
And will probably lead to an explosion of attributes.
I could definitely see the side of someone who says: who cares, I'm just going to remove the warnings in 95% of the cases, and its dammit from then on
Joe4evr
@Joe4evr
+1 for dependent variables, I got a scenario pretty much identical to Task.Status, but I also understand it'd be a huge cost on top of what they've already done for NRTs
Andrew Hall
@ryzngard
@jnm2 did you figure out the delegate scenario? You still probably want FooDelegate<string>(out T? t) with some annotations regarding when it doesn't return null
CyrusNajmabadi
@CyrusNajmabadi
image.png
there is so much about windows that confuses me at times
'moving 37k items'. yup, that's about right
"0 remaining items" hrmmm
"50.8gb remaining" hrmm
4% complete. hrmmm
i honestly can't really form any sort of mental model around this operation
Joseph Musser
@jnm2
@CyrusNajmabadi I discovered that robocopy is very fast
like on the order of 10-100 times faster
This was from one M.2 SSD to another on the same mobo
@ryzngard I'll construct a more complete example so that I'm not accidentally stringing you along. Your advice is much appreciated!
Joseph Musser
@jnm2
@ryzngard Here it is: dotnet/roslyn#38191
@YairHalberstadt the amount of effort it took to post dotnet/roslyn#38191 is why I'm not jumping up and down to spam them with all the situations I encountered :D
Plus, maybe I learn something in this issue that impacts some of the other situations
Jason Bock
@JasonBock
Analyzer question: Should I always check the given node from RegisterSyntaxNodeAction for diagnostics? In other words, will the node given to the callback method be diagnostic-free, or is it possible to get nodes that have diagnostics/issues?
Johan Larsson
@JohanLarsson
there is API for that iirc
what problem are you solving btw?
or is it possible to get nodes that have diagnostics/issues
my guess is that relying on this can be fragile due to async
Alireza Habibi
@alrz
anyone knows why System.Runtime.CompilerSerivces.ITuple is not available in roslyn?
Johan Larsson
@JohanLarsson
how do you mean available?
Alireza Habibi
@alrz
couldn't reference it (compilers says that there's no such type)
I guess some package is not up-to-update?
Johan Larsson
@JohanLarsson
you did mandatory close vs & git clean right? I've never used that type for anything so dunno any specifics.
Jason Bock
@JasonBock
@JohanLarsson I know that you can check the node for diagnostics, that's not my question (if that's what you're getting at). What I'm saying is, if the given node has diagnostics, then I (probably) don't want to do any analysis. But does the analyzer infrastructure only give you nodes when they don't have issues, or does it give it to you regardless? I have no problem adding an if to check for diagnostics, but if I don't have to do that, then I'd rather not add unnecessary code.
Johan Larsson
@JohanLarsson
I have never checked that and never seen any issues. WHat problem are you solving with the if?
Jason Bock
@JasonBock
@JohanLarsson I don't know if I'm trying to solve a problem or not.
All I'm trying to get an answer to is if the node given to the analyzer will not have diagnostics, or is that possible?
Johan Larsson
@JohanLarsson
ok, we wait for cyrus then I was just curious, not trying to be a dikk
guess we could fire up vsix and get answers before cyrus
Joseph Musser
@jnm2
@JasonBock I'm pretty sure diagnostics are all run in isolation and in parallel
By design, not being able to depend on each other
@CyrusNajmabadi can you confirm?
CyrusNajmabadi
@CyrusNajmabadi
In other words, will the node given to the callback method be diagnostic-free, or is it possible to get nodes that have diagnostics/issues?
You can check if you want. It's up to you
When you register a syntax node action, you will be called back for all nodes of that type. No matter what
All I'm trying to get an answer to is if the node given to the analyzer will not have diagnostics, or is that possible?
There is no support for this concept.
Note: diagnostics are only loosely related to nodes. For example you can have diagnostics with no relation to any specific node.
Andrew Hall
@ryzngard

@ryzngard Here it is: dotnet/roslyn#38191

Thanks :)

Johan Larsson
@JohanLarsson
When does Document.TryGetSemanticModel return true?
Yair Halberstadt
@YairHalberstadt
After it completes ;-)
Johan Larsson
@JohanLarsson
:)