MetaTalk back button anomaly

August 7, 2017 2:04 PM

So I'm reading an active MetaTalk thread. The "[N] new comments, show" message appears. I click on "show" and get the new messages. I click on a link in one of the new comments and go to a different page. I click on the browser Back button or press the backspace key. I end up back in the MetaTalk page, but without the new comments. Some short time later, the "[N] new comments, show" message appears again.

If I go to an entirely different web page and hit the back button, the same thing occurs, I get the MetaTalk page without the new comments.

I haven't done a lot of experimentation. I tried it once on the blue and came back to the page with the new comments.

I'm running Win7 and the latest Firefox on a fairly elderly HP laptop.

I categorized this in Bugs because there was no better choice, but I'm not sure it's a bug.

I had to check the "My request requires community input ..." box to get this to post. I don't actually think it requires community input, but I'd be interested to know if other MeFites have seen this behavior.

posted by Bruce H. to Bugs at 2:04 PM (13 comments total)

Yep, that's I think the normal behavior on MeFi, which is to say it's not a bug so much as a design issue with how page loading and the "load new comments" feature is implemented. I don't recall whether we determined there was an appropriately light-weight fix for it that's lurking on the TODO list or if it had looked like more of a pain in the ass than that.

I tend to give a thread a hard refresh now and then if I'm keeping the tab open all day or multiple days, since there's other details that can also end up out of sync over time (in particular, comment deletions).
posted by cortex (staff) at 2:07 PM on August 7, 2017 [1 favorite]

OK, but why does it work differently on MetaFilter vs. MetaTalk? I just tried it again on the FPP about the private street in SF, navigated to Tamara Keel's blog, and when I came back, it was to the page with the new comments.
posted by Bruce H. at 9:35 PM on August 7, 2017

OK, but why does it work differently on MetaFilter vs. MetaTalk?

It doesn't work differently for me. No matter which subsite I'm viewing, "Load new comments" seems to push an anchor #inline-whatever onto the history, and going back to that does not include all comments. I've always just worked around it by doing a full page refresh when I want to go back (and avoiding the issue by opening everything in new tabs when possible).
posted by a snickering nuthatch at 10:36 PM on August 7, 2017 [2 favorites]

Or maybe that's been fixed since I last tried it and my workarounds are no longer necessary... I could've sworn I've seen this behavior recently in the megathreads at least, though.
posted by a snickering nuthatch at 10:45 PM on August 7, 2017

Okay, I think I've narrowed down the behavior I'm observing (on the blue). Once I click "load new comments" once, the url contains #inline-[some number] and the browser appends the first batch of new comments onto the page.

If "load new comments" appears again (so there's a second batch of new comments waiting), but I do *not* click it but instead navigate somewhere else, then try to navigate back, the browser returns me to the page with only the first batch of new comments and not the second.

If I had never clicked on "load new comments", such that the url does not include #inline-[some number], then navigating away and then navigating back causes the page to load in full, including any comments that were waiting on the server but my browser hadn't loaded before.
posted by a snickering nuthatch at 11:47 PM on August 7, 2017 [1 favorite]

This is probably a good place as any to note that the "(X new)" count of unread comments next to posts on the main pages of metafilter never seems to reflect whether I've read those comments already or not. I haven't narrowed it down, but I think it's recording (X new) since I last clicked on the main page link & isn't noticing other visits to that page, or the times I've clicked on the "more comments" link when it appears.

Is this expected (if maybe not ideal?) behaviour or is it peculiar to me? Am I misunderstanding the intent of the "new" in "(X new)"? Is it new relative to some other metric than the last time I opened that post page?
posted by pharm at 3:44 AM on August 8, 2017

Is this expected (if maybe not ideal?) behaviour or is it peculiar to me?

I'm pretty sure this is normal; at least, I know it happens to me, and I seem to recall there being a MeTa way back when that discussed in some detail how the "X new" was estimated and that it would not be exact.
posted by solotoro at 6:51 AM on August 8, 2017 [1 favorite]

Yeah, the front page "x new" functionality is a much older, less precise estimate based on browser sessions. It's more or less on target when you're coming back to the site after hours/days away but not much use for more incremental scenarios where you haven't really been away from the site for the last bit. It's something we may eventually rework but historically it's been and remains more of a rough yardstick than anything.
posted by cortex (staff) at 7:12 AM on August 8, 2017 [1 favorite]

The relationship between the front page "x new" display and reality is so weird and unpredictable to me as to make that feature worse than useless. If reworking it isn't feasible I'd be in favor of just dropping it. It obfuscates more than it elucidates, if you ask me.
posted by Anticipation Of A New Lover's Arrival, The at 4:07 PM on August 8, 2017 [2 favorites]

It's a wibbly-wobbly timey-wimey thing.
posted by XMLicious at 9:31 AM on August 9, 2017 [2 favorites]

I think the problem is that the site is using the HTML5 history API to override the current entry.
This seems like an unnecessary feature.
posted by Tobu at 11:04 AM on August 9, 2017

It's a real problem on mobile browsers (which tend to toss out pages liberally to save memory), but a work-around is to click on the time of the most recent comment you've read and then when you notice the page has reloaded, hit refresh.

(If it's different on metafilter and metatalk, I haven't noticed)
posted by Yowser at 8:26 PM on August 9, 2017

I guess that doesn't really fix hitting the back button, but it's the same issue.
posted by Yowser at 8:27 PM on August 9, 2017