[en.topic_count_all] Click to show.
-
Oh dear, something isn't substituting right:
I've been seeing this on and off for a while, mostly on, when I go away from my keyboard and some new posts come in. I'm guessing based on past experience that there's supposed to be some kind of number in there to say how many new/updated threads there are, but right now we've got Fail.
-
Yeah, that's been going on for like a month.
-
I think it has already been reported on meta.d, and it’s a CANTREPRODUCE WONTFIX JUSTUSEF5.
-
Refreshing is a workaround, not a fix. A fix would be to take the unsubstituted bit out of the message. In UI terms, it can be better to not try at all than to try and fail.
UIs are not known for being brave.
-
Well, in the Discourse Design Handbook, the existence of a workaround seems to imply DONTNEEDTOFIXRIGHTNOW or possibly DONTNEEDTOFIXLIKEVER (but very definitively it implies DONTNEEDTOFIXFORV1).
Filed under: [I was going to link to an example, but couldn't decide which one to pick.](http://what.thedailywtf.com/t/inserting-links-after-hr-fails/825)
-
-
FYI, it's been seen on meta.d. They don't seem to have a repro though...
https://meta.discourse.org/t/en-topic-count-all-click-to-show/18995
-
The problem may be that they're trying to repro with the assumption that Discourse operates in a deterministic, cause/effect-driven way.
Filed under: meta.discoursistency
-
A fix would be to take the unsubstituted bit out of the message. In UI terms, it can be better to not try at all than to try and fail.
Or to just have an error handler that makes sure the thing they want to print is a number... if it's not, then just print the "click to show" bit only.
Filed Under: Error handlers in Dicsourse? Clearly doing it wrong.
-
The problem may be that they're trying to repro with the assumption that Discourse operates in a deterministic, cause/effect-driven way.
Exactly. I'm almost certain this is another race-condition induced bug, where that a shared global variable somewhere is getting trashed by a different ajax loading while the first was in the middle of processing.
-
certain this is another race-condition induced bug, where that a shared global variable somewhere is getting trashed by a different ajax loading while the first was in the middle of processing
Yeah, if only I could get away with that where I work.
- Build multi-threaded application without thread safety.
- Demand consistent repos from users.
- ??????
- Profit!
-
Or a call to see if there are new posts succeeds, but the call to see how many there are fails (shouldn't it be one call? but I'm buttuming). The replacement text isn't replaced, and just confuses the user instead of showing a useful message, such as "there are new topics, but we have no idea how many because something broke."
-
3. ??????
Require enterprise support
Nice, discourse.
@Arantor: Days since bug found in Discourse - 0
-
Indeed it is. That's today's quota met. Keep up the good work.
-
That's old, it's a Markdown bug, and the Discourse team thinks it's not actually a bug.
I encountered it when someone asked a question requiring a numeric answer, and Discourse changed my answer to one. Something like "How many days are in a week?"
- Holy smokes it actually displayed 7 this time! (erm..now it says 1 again...idk)
-
-
Anyone remember the URL syntax to show a raw post? I can't find it and I'd like to add a link AND put the link in my bio.
-
Anyone remember the URL syntax to show a raw post? I can't find it and I'd like to add a link AND put the link in my bio.
Just replace the "t/post-title-whatever" with "raw". @PJH will be around in a bit to scold you with a link to his FAQ.
-
then he can add the link to the faq to his useful links bio
-
Win/win situation.
Especially if there's a userscript that can offer it that can also be linked to.
-
Bug posting to the bug topic:
Back to the original bug, this has been pretty reproducible for me, starting with the v1 update. Before that, it was sporadic; now it's almost consistent.
New post arrives while looking at the topic list, usually updates correctly. Read a topic; new post arrives in another topic. Back to the topic list, broken. And when it's broken, it's really broken. Not just the en.topic_count_all, but the new/unread post count indicators and the activity sorting fail to update.
-
I never really got why they didn't make the listings view be a real different page than the individual topic views. That's where so much of this sort of weird stuff comes from.
-
This was a tough one, we should have it fixed tomorrow.
https://meta.discourse.org/t/en-topic-count-all-click-to-show/18995/10
-
explains why certain people saw it so much and other never did, hitting back is a user-browsing-style thing.
-
Except I mostly saw it on mobile, where I never hit "back".
Unless there's multiple ways to hit that same point of failure. Guess we'll find out tomorrow if the forums are updated.
-
Also interesting about the Discourse.TopicList tracker "not being triggered" thing, because it doesn't just affect the topic list page, i see that "en.xxxxxxxxxxx" in a few other places too. Like the "X Replies" button.
Also, I pretty much never use the back button but have had this error happen to me as well. It just hasn't been an increase since v1 for me, it's been about the same as ever.
-
ok - reproduced it here using back.
wonder if there are other navigational subtleties involving the js overridden links that can cause it to happen on other pages as well.
-
hitting back is a user-browsing-style thing
I use back all the time. I've been doing it since the days Mosaic ruled the Web, and I'm not about to stop. Jeff already told me I'm Doing It Wrong™. Tough. If a web app can't handle normal navigation; that's the app's fault, not mine.
And yes, since v1, I see this error just about every time I do.
-
I'd look at the report but I'm getting 500 errors from meta.d.
-
-
that was my bad, sorry
-
Anyone remember the URL syntax to show a raw post? I can't find it and I'd like to add a link AND put the link in my bio.
http://what.thedailywtf.com/t/viewing-source/780
http://what.thedailywtf.com/t/obscure-urls/2086
then he can add the link to the faq to his useful links bio
-
that is not mott's bio, no help at all!
-
Question from the peanut gallery: how does a 'corrupted link' produce a 500 from a server? There is no legitimate circumstance where a 500 should ever be intentionally thrown for a URL. Bug?
-
Question from the peanut gallery: how does a 'corrupted link' produce a 500 from a server? There is no legitimate circumstance where a 500 should ever be intentionally thrown for a URL. Bug?
Even things like http://what.thedailywtf.com/?a result in a blank page with a 500 status code, so I’d guess it’s Discourse way to tell that it does not like a request.
If only there was an HTTP status code for a bad request, they could probably use it... Wait...
-
There's plenty of choice.
400 is the obvious, for syntactically invalid request. 404 wouldn't be entirely inappropriate, neither would 403. Heck, 405 wouldn't be entirely inappropriate. Nor 410.
Fuck it, I'd even encourage 418 to be used there.
5xx codes are for where the server is not capable of replying, as opposed to 4xx where there is something wrong with the request. The request is broken in this case; it's not the server's fault, so 4xx not 5xx errors should be used.
FFS they could even invent a new non-standard number in the 4xx range and that would be an improvement semantically.
STOP BREAKING THE FUCKING WEB, DISCOURSE.
-
Had this error on the train this morning when I lost my connection with chrome (latest) to the net.
-
Seems Jeff doesn't know about the history api ...
-
Fuck it, I'd even encourage 418 to be used there.
Any Discourse response is anything but short, even though it may be stout.
-
To put that in true Discourse terms,
400 and the other 4** codes are all variants of “You're Doing It Wrong”.
500 and the other 5** codes are all variants of “Ooops! I did it again”.(I guess we need 900 codes for indicating both simultaneously.)
-
In which case it's still wrong
-
yep, its a bug, at worst it should be a 404 there
-
Filed & resolved
https://bitbucket.org/masamunewos/discoursebugs/issue/28/entopic_count_all-click-to-show
