My stackoverflow question was closed so here's a blog post about CoreWCF

Same here.

I have a rep that is based almost entirely on questions, not answers. I learned to ask questions fairly well (to the point I seldom get answers -there's a price to pay, for questions that are very specific).

In some cases, the question is a basic one, and doesn't need a code listing and sample project. It's still a perfectly valid, pertinent, thoughtful question, but not very verbose.

Those questions almost always get closed.

I have found that asking LLMs doesn't always get me the best answer, but I get an answer. In some cases, I can have an iterative refinement, where I keep adjusting the question, until I get a useful answer.

I've never gotten code from SO, that I can use without modification, but I have gotten some great answers, over the years, and have expressed gratitude and respect.

I have gotten quite annoyed with the "attitude" that is often expressed. There's no doubt that folks who ask questions, are considered "lesser beings" on SO. Just look at the question-to-answer ratio of the high-score individuals. Weird attitude, for a site that is pretty much completely reliant on questions.

Basically, I have just given up on SO, and have found LLMs to give me what I used to get from it.

In my opinion, they have killed SO.

I’d appreciate if somebody more familiar with SO would verify this, but I believe there’s some low constant number of close votes required to close something, and this doesn’t adapt to how many people are voting or to positive signals. Because there’s an error rate in all things, this naturally means that things are wrongly closed all the time, especially content that’s viewed a lot and not fought for.

And the irony here is that much of what LLMs know is from training on StackOverflow.

I had stopped interacting with SO as well, though the fact didn't really cross my mind. I had a few popular answers that got be enough points to be a moderator. The experience there is similar--possibly worse because the idiocy is unveiled. In the past I've often argued to reopen questions, sometimes even making edits to make it more agreeable to other mods.

It's common for those to get shouted down based on some policy or other bureaucratic nonsense by those who have no idea what the question is actually about. The problem could be that many of those who don't do, moderate. It attracts different sorts of people than those that are actually working with the things being discussed.

The moderators were elected. What should StackOverflow have done, held a vote of no confidence? Given them less power; make moderation more democratic?

Somehow I never found the StackOverflow game to be worth playing.

In retrospect it is a case study of a particular enshittification scenario: "benign neglect" Back when they published a data dump I had a project on my speculative list to clean up their database, take only the best answers, etc. For python, the numerous Python 2 examples

   print "something"

   print("something")

I stopped flagging things on SO when some of my flags were deemed unhelpful. They clearly weren't.

just a few months ago they closed a question of mine that I posted in 2010 (!), which in the meanwhile had gathered more than 1000 votes, nearly one million views, and 20 or so answers. I get it that it does not meet their most recent criteria, but closing a question after 15 years telling me to edit my question and read the comments on how it could be improved (there were none) sounds tone deaf and unnecessarily bureaucratic

Hey mate is it your post? I did glanced at it and it does not stop because server is not notified that client is not there.

Or at least that is my guess, since I stopped working with WCF about 2016 probably.

Anyway in newer version of .net you CancellationToken everywhere what would do exactly that: tell your server that client disconnected. That would be my first try on fixing it.

Use token that is sent via HTTP implementation to the endpoint, pass it to your stream and when it is cancelled, end the stream. Stream ends, endpoint finishes, not CPU load.

I don't see anything wrong with your SO question (I am a long time contributor), and don't see why it would have been closed.

I will say, this is a level of question that is too sophisticated for SO, and likely will only have an answer once you figure it out and go back and answer your question.

Are you confident the code is the issue--have you repro'd it consistently with different versions of .NET? What about reproing on different machines? Locally?

Honestly, I agree with it not being a good fit for a Q&A site. It's a debugging problem, probably needing a discussion, and might even not be of any use to others being that "high cpu" is kinda vague. Seems better suited for a bug report / issue tracker of the relevant library.

for what it's worth, I submitted the question as a candidate for reopening

> Please feel free to point out more ways in which I screwed up asking my SO question.

With pleasure! SO is definitely more of a distinct Q&A site and not a discursive, open-ended collaborate and problem-solve site.

I fully expect nobody in this comment section to care about the CoreWCF content. (I don't even know offhand what that is.) In my experience, people love talking about Stack Overflow in places that are about programming but aren't Stack Overflow, so.

(Edit: it seems people do care about CoreWCF ITT. That's nice to see.)

> Screaming into the void of the blogosphere is catharsis for getting my SO question closed.

That's fine. Almost everyone who comes to SO, in my experience, has a fundamentally wrong idea about how the site is intended to work. That includes people who don't have a question and only want to post answers. Unfortunately, it's difficult to explain because people find the model unintuitive - the UI affords using the place just like many others, even though the site was created exactly to get away from frustrations caused by older models (https://meta.stackexchange.com/questions/92107). And the real objective is a synthesis of many not-always-compatible ideas (https://meta.stackoverflow.com/questions/254770). My personal sense is that the community didn't really get a handle on "what SO is" until around the time that new question volume peaked (way back in 2014).

Even then, people can hang around for years and not really get it (e.g. https://meta.stackoverflow.com/questions/427224) - in large part because the policies have been inconsistently applied on a volunteer basis, and the people who are allowed to e.g. cast close votes are vastly outnumbered.

We generally don't care about people not liking the Stack Overflow model while discussing it off-site. There's far too much of that to worry about. But that doesn't mean we'll change to accommodate everyone else. The entire point is to provide something that isn't available everywhere you look: a polished artifact, an organized repository of commonly-needed, high-quality answers to clear, focused, practical questions.

Do we accomplish that goal? Hell no, not by a long shot. But there are some real gems in there - and a few of them have millions of views. And as the rate of new questions slows, users who put on the "curator" hat become able to keep on top of the incoming queue, filter through for what's of value (and not a duplicate), and even turn attention towards the old Q&A to improve it (incidentally, a lot of that work is rounding up old duplicates that went unnoticed).

> I had forgotten that any external links are a big no-no in SO land, so my question immediately attracted 2 close votes.

The problem isn't simply including an external link (we'll happily just edit those out if they aren't necessary). The problem occurs when a question appears to depend upon the externally linked content. We can't accept that (https://meta.stackoverflow.com/questions/254428) because of link rot and licensing issues (someone who wants to answer you often needs to be able to cite the code; posting on-site automatically licenses the content appropriately, per the terms of service) but mainly because of scope - a question that's suitable for the Stack Exchange format would fit neatly within the actual question text.

We don't want to do detailed analysis of the problem you encountered, even if we're capable of it, because questions are for everyone. They need to be able to reflect a problem that other people could a) have; b) plausibly search for; and c) recognize if they found it. Answers to a question need to make sense in general to people who would ask - not just in the specific context of one person's original problem. In short, we want a question, not a problem - and extracting a proper question starts with (https://meta.stackoverflow.com/questions/261592) your own analysis.

"How do I do X?" questions are usually much easier to ask in the format, and are very valuable and can end up very well regarded, even when they're on very basic topics. But "what went wrong with Y code?" is not fully refined. What we're really looking for is more like "why does Y' code construct do Z?" - where the specific, exact cause of failure (https://stackoverflow.com/help/minimal-reproducible-example) is extracted from your own debugging session (along with reproducing input and actual vs expected output).

> Two days later my question got it's third vote for closure, and remains unanswered and now closed forever.

This is literally not how Stack Overflow works. The OP has at least (https://stackoverflow.com/help/auto-deleted-questions) 9 days to fix the question and nominate it for reopening until it gets "deleted"; but even then it's a soft deletion (delisting) which is still reversible - you can find the question from your personal listing (https://stackoverflow.com/users/deleted-questions/current while logged in; or replace 'current' with your user ID), edit and nominate for undeletion.

The established policy is that we intentionally close questions that don't meet standards (https://meta.stackoverflow.com/questions/417476) as quickly as possible (https://meta.stackoverflow.com/questions/260263). The main point of this is to prevent the sort of people (notice that https://meta.stackoverflow.com/questions/271684 is over 10 years old; and the original complaint https://meta.stackexchange.com/questions/9731 is from before the official launch, during the private beta) who would otherwise hang out on a traditional discussion forum 12 hours a day from trying to read the OP's mind, repost the same basic explanation of the same basic idea dozens of times, etc.

(Unfortunately, the incentive system is completely broken - https://meta.stackexchange.com/questions/387356 - and the company's interests are not aligned with the community, so this is a losing battle.)

And, in fact, your question has been reopened, as of about 3/4 of an hour after your comment that I'm replying to. Stack Overflow is not at all immune to external pressure - after all, many regulars there are also on HN and other usual-suspect sites.

It also looks like your edits have actually improved the question. In particular, adding in a definite conclusion from your profiling attempt.

(We understand that a lot of people in a situation like yours wouldn't necessarily know how to use a profiler and wouldn't necessarily be able to come up with a theory about what's wrong. That isn't our problem. We aren't offering tech support. It's a bitter pill for almost everyone, but Stack Overflow by design is not there to make your code work. It's there to answer questions that arise during your attempt. And a question like yours, properly refined, can help those other people.)

Likely this summarises the shift:

https://openai.com/index/api-partnership-with-stack-overflow...

Thanks for trying to help.

On the client side, randomStream.Close will get called when it's disposed.

On the server side, I'm not sure what I could put into an overriden Close that wouldn't just be base.Close()? RandomStream itself doesn't own any resources that need cleaning up.

I could force WCF to use Session mode, and then add flow-control through a side-channel, so other messages could prepare the stream to internally buffer and then rewrite in requested chunks?

But at that point I might as well just use an apprpriately sized GetRandomBlock(ValueWithSequence[]), and chunk requests that way and abandon using a stream for this at all.

I'll have an experiment with that approach to try to find the best buffer size and whether streaming the buffer actually helps vs just having it as the message and letting WCF control the sending.

Stupid idea: throttle stream by putting Sleep in Read

I didn't, but I've not had much previous success with these kinds of issues.

Just for the sake of it, I've now tried pasting this whole blog into claude.

It has some strange suggestions, including many things that don't work, such as adding to the client:

                // Important: Properly close the stream
                randomStream.Close();

> This method calls Dispose, specifying true to release all resources. You do not have to specifically call the Close method. Instead, ensure that every Stream object is properly disposed.

My stream is properly disposed.

Claude also sticks in things like:

                // Simulate some work or add a small delay to prevent CPU spinning
                Thread.Sleep(1);

On the client side it not only rewrites my reading to read multiple times, but adds in:

                    await Task.Delay(500, cts.Token); // Small delay between reads

The only thing that was interesting was creaitng a linked cancellation token source on the service to pass to the underlying stream and cancelling it on server shutdown.

That's a useful thing to keep in mind to help with helping to shutdown the services, but doesn't actually address the issue for a server you want to keep running.

It does also add send/receive timeouts too, and they're also worth keeping in mind, but that doesn't seem like a good mechanism for dealing with this issue. If anything, it would just mask the issue by having it write to the stream for the duration of the timeout period instead, which if short could cause a problem like this to go unnoticed until it's actually under more load.

Thanks. In my experience the CoreWCF team are fairly active, but I didn't want to bother them until I can understand if this is a problem with CoreWCF.

This might also be a problem WCF client, which is maintained by others elsewhere in a different repo: https://github.com/dotnet/wcf for the nuget package version.

But this might just be how WCF is designed. I'll try a version of this within .NET Framework, but even that might change depending on whether it's via IIS or started via ASP.NET Core, and whether it uses the built in System.ServiceModel or the nuget version.

( You can probably tell I'm a bit frustrated with MS for making a bit of a mess in the way they hurried away from .NET Framework especially with respect to WCF. )

... and https://github.com/CoreWCF/CoreWCF/discussions

What resources do you think should or could be handled in an overriden close method on RandomService?

It'll be calling base.Close(), and doing what else?