Comments Since Last Visit, Reloaded, Augmented, Installed, In Two Steps


CK MacLeod

WordPresser: Writing since ancient times, blogging, e-commercing, and site installing-designing-maintaining since 2001.

Related Post Roulette

50 Responses

  1. Avatar Nevermoor says:


    I’m going to give it a few days of real use to see if I have other thoughts, but otherwise this is excellent work.Report

    • Avatar nevermoor says:

      Recognizing I’m probably way too invested at this point, a few thoughts:

      1. The clickable buttons are showing a thin rectangle instead of an arrow. Maybe an invalid character reference?

      2. It’s not clear to me exactly how the buttons work. What I expected was that the button at any particular comment would point at the subsequent unread comment. I’m getting different behavior:

      Assume 10 unread nested comments. If I have 1-3 on the screen, and click the button next to #3, it scrolls from 1 to 2. If I instead scroll down so that 4-6 are on the screen, and click the button next to #6, it seems to scroll to #5 (i.e. “next after top of screen” logic). I think my expected behavior would be more intuitive and functional, but you’re the designer.

      3. It appears that the last button points to the top of the comment thread. Might prefer the first NSLV comment, but there is (of course) a button right there so we are only talking an extra click.

      4. The only UI issue I see with putting a button at the bottom right of a comment (especially with the functionality in #2) is that it breaks focus when reading. This might be because I’m used to scrolling through threads with a keypress, but I find myself pulled away from text to find the button I need to click when jumping to the next group of NSLV comments. I’m not sure what the solution is (other than parallel shortcut keys),Report

      • Avatar CK MacLeod says:

        Thanks for the feedback, and please don’t apologize for taking an interest: You’re the only one who has.

        1. Are you able to provide a screenshot? Also, what browser/os?

        2. Again, knowing browser/OS combo would be helpful, also if you view the threads at some unusual magnification level. In general, clicking the button should take you to the comment with the next one. With long comments or in other ways oddly vertically situated comments, it might take you a second click, or an extra scroll, to get to the clicker.

        I haven’t noticed the behavior you found about being on 3, and going to 1 or 2. If you could point me in addition the specific thread where you encountered that sequence, that might also be helpful.

        3. I found from my own experience that going to the top of the comment thread to resume the cycle was more convenient, since at that point I might very well want to show only new or refresh as start the cycle again with the first new comment.

        4. Noted. But I think you’re right about your being used to having a keypress alternative.Report

        • Avatar nevermoor says:

          2: normal magnification (Windows/Chrome) Was playing around with buttons in this thread. Maybe it’s an interplay issue with your comment snaking, as I happened to be looking at reverse-nested comments at the time I wrote that, but if I have the screen centered on North’s “I’m gonna check it out!” with the comment from Glyph above entirely in-screen (and both unread), clicking the button next to North’s comment scrolls to the top of that comment rather than the next one.

          3: fair explanation.

          4: I definitely recognize there’s a limit to which your solution should cater to my unusual preferences.Report

          • Avatar CK MacLeod says:

            on #2, that sounds a little less unexpected/problematic than the earlier description.

            If you start with the GTN BUTTON up at the top, then click your way through clicker by clicker, you should ideally bring each comment in succession to the top of the screen, then reach the last one, click, and be returned to the top of the comment section.

            If you jump ahead, or start the sequence by clicker rather than by button, then it there might be a slight hitch when the code calculates its target: as an offset from the top. It will end up finding the prior clicker and scrolling to show the associated comment as “next unread.” BUT next click it should continue proceeding down the list, and, only at the last, targetless clicker, scroll back to top.

            So, if there are comments 1 through 10, and you have come to a screen showing you 4 and 5, and click 5, the code first finds the 5 clicker as the “next target,” and then should go to 6, as new next target, on next click.

            In my use, I found that kind of within “bounds of the expectable,” if you will. But clicking on 4, then 5, then getting stuck in some kind of loop that never got you to 6 and beyond – that I would find disturbing.Report

            • Avatar nevermoor says:

              To be clear: there was never a loop. I would just expect that no matter what my screen is showing clicking “next” at a comment will take me to the NSLV comment after that.

              People won’t just click through all the comments (in part for the UI reason I noted above) but will likely scroll through unread chunks until they need the button to jump ahead. It’s strange, under that use condition, to click the button and be jumped back up thread.Report

              • Avatar CK MacLeod says:

                Next time you run across the problem, please take a very close look at what’s happening that you find confusing. It’s hard enough to state clearly, but I think that wherever you are and whenever you click, unless you’re on the last new comment, you’ll always go down. I figured users would find that sufficiently in bounds (and just keep clicking!), but I agree it’s not toadly optimal.

                So, first I’ll replace the character code with image icons,, and then I’ll see if I can jigger the jQuery to make things smoother for out-of-sequence/midthread clickers. It may be simple, or it may require more complicated offset calculation and array counting. Either way, it may facilitate some other more advanced indications (like producing down icons when down-clicking, and an up or recycle icon when you’re on the last new comment in the list – but no promises!).

                Thanks as ever for getting down into the details on this.Report

              • Avatar CK MacLeod says:

                You know one thing I forgot to ask: I’ve already replaced the arrows character in the top button with an image icon version, but I wonder if it was showing for you while the side-clicker wasn’t.

                Will get to the side-clicker shortly. EDIT Er – later…Report

              • Avatar nevermoor says:

                Arrows now displaying correctly.

                The time it goes up is when I’ve read a few comments on the screen and clicked the button next to the last of ’em. Makes the screen scroll backwards/up every time on my setup.Report

      • Avatar Michael Cain says:

        The clickable buttons are showing a thin rectangle instead of an arrow. Maybe an invalid character reference?

        It renders properly on my screen. This does raise the philosophical question as to what makes an invalid character reference. The character code is a valid Unicode glyph; OTOH, lots of fonts fail to define the arrow glyphs. This is part of why I’ve made the Noto fonts my default — Google has committed to defining every glyph at least up through Unicode 6.2 for them. The other reason is that with some spacing tweaks, the Noto Serif font is quite attractive.Report

        • Avatar CK MacLeod says:

          I’d – obviously wrongly! – expected virtually all browsers to handle the character correctly. I may just use an image (or two) instead, though I’ve tried applying a very generic serifed font stack specifically for the clicker arrow (both top and side), as an experiment just to see whether it fixes the problem for nevermoor for now.Report

          • Avatar nevermoor says:

            Now I have a square empty box instead of a thin one. (Windows/Chrome)

            It’s trying to render the glyph in Georgia. (font-family: Georgia, Times, ‘Times New Roman’, serif;)Report

            • Avatar CK MacLeod says:

              Odd – have no idea why it would be I think I’ll just replace it with image icons. On Windows 7/Chrome, I’ve had no problem either way, nor on Android. Have you tried a cache flush? Seems this wasn’t a problem for you when we were discussing the earlier versions on the Developing blog?Report

              • Avatar nevermoor says:

                True, it wasn’t a problem there. Just did a hard reload and commenting to get an NSLV.Report

              • Avatar nevermoor says:

                Not fixed.Report

              • Avatar CK MacLeod says:

                Switched to decimal from hexadecimal code just on off chance it works for you. Decimal is, in my observation, more common in HTML. Had only used hex because in an early draft it was part of a CSS “content” statement, and for those you use escaped hex. One difference that only the obsessively observant would notice is that this is a different double arrow character than the one I was using on Developing: It’s down then up, rather than up than down, though the difference in the code was only an F rather than a C.Report

              • Avatar Michael Cain says:

                I’ve seen versions of fonts that include one glyph but not the other. To some extent, fonts can be the kind of DLL hell that some of us used to suffer through — no version numbers, upgrades break as much as they fix, etc. This is the kind of problem that has led to abominations like Font Awesome.Report

              • Avatar CK MacLeod says:

                Will teach me to distrust the code for the also-ran characters anyway, though I still lack a good guess why my Chrome is getting what nevermoor’s Chrome isn’t, and why it worked OK for him on Developing…, but not hereabouts. If it’s not a problem 100% unique to this site, it’s worth encountering the exception early rather than late.Report

              • Avatar nevermoor says:

                Yep. Color me mystified too. Image is the right solution, though, especially if you see this as a shareable WP plugin (which you should, it’s great work!)Report

        • Avatar Michael Cain says:

          Need another new comment to look at something.Report

  2. Avatar Jaybird says:

    Can we have the Greek Alphabet? Or is that crazy talk?Report

    • Avatar CK MacLeod says:

      All things are possible in this best of all possible worlds, Jaybird.

      I’ve just added CSS support for Greek. So, if you can type Greek characters, you would wrap them in a span tag, like so (experiment #1):

      ??? ???? ????? ??????? ??. ?? ??? ???? ???????, ?? ??? ?????? ?????????, ??? ?? ??????? ????????. ?? ??? ????? ????????, ??? ?????? ??????? ??. ?? ??? ?????? ??????? ????????. ??? ???? ?????? ??. ?? ??? ????? ?????, ?? ??????? ???????? ???????? ???, ??? ?? ????? ????????.

      hmmm… still working on it – didn’t turn out to be as simple as indicated.Report

      • Avatar Brandon Berg says:

        Your problem’s not on the font side. It’s on the input side. The input plugin is converting all non-Latin characters to question marks. I saved this page as a local HTML file, edited the source to change the question marks in your comment to Greek and Asian characters, then loaded it in my browser, and they displayed just fine. If you can get the input plugin to stop eating non-Latin characters, everything should be fine. I’m guessing that there’s a simple setting to do this.Report

        • Avatar Brandon Berg says:

          @ck-macleod Not sure if this is the issue, but it may be that your SQL tables use a Latin character set, in which case the solution (or at least one part of it) might be to change the collation.Report

          • Avatar CK MacLeod says:

            Didn’t want to start futzing with the database directly on a live site – especially changing the collation on a site with archives as large as this one – not when there are preferred ways to handle the issue, and there are warnings about unexpected results and the amount of time it may take to perform a conversion.

            I’ll do some experimenting in the development environment version of the site later.

            As I noted above, if we REALLY had an interest in doing lotso Greek, we could load fonts with the appropriate character sets (they come in different flavors for classical vs modern Greek, for instance). I’m not sure what you had in mind in the previous comment by “input plugin.” There is a WordPress setting for the language of your site, but there’s no built-in WP setting for making your English language site add Greek in a way that will render as Greek on browsers generally – and limited information on the subject beyond what I’ve already referred to.Report

            • Avatar CK MacLeod says:

              Also, high φ.

              Psych! (or rather ????)Report

            • Avatar Brandon Berg says:

              Fair enough, although I wouldn’t expect it to take that long. The vast majority of characters on this site are ASCII, and ASCII characters have exactly the same encoding in Latin and UTF-8 encodings.

              Is that what you did to get Greek characters working on the test blog? I saw that they actually appeared in the source, so I guess that means they’re in the database?

              I’m not sure what you had in mind in the previous comment by “input plugin.”

              Nothing, really. I don’t know much about WordPress, so I assumed there was some plugin that handled comment submission.Report

              • Avatar CK MacLeod says:

                Well, what appears in the source is not necessarily what appears in the db, but in this case I think it is, and I think (will check!) that all of the stuff converted to question marks above is also whatever it was before being questionmarkified.

                As for WP, there’re lots of things that go into producing this wonderful comment thread. “Plugin” usually refers to an application or set of functions or whatever you want to call it that can be optionally added to the “Core.”

                As for high phi’s – here’s some inline versions.



              • Avatar Brandon Berg says:

                I know how to do the superscript, but when I try to add style to a span tag, it gets taken out, probably because allowing commenters to add arbitrary style directives could mess stuff up pretty badly. There’s an HTML sup tag, but I’m not sure if that’s allowed: testing.

                Edit: Nope. Gets eaten up.Report

              • Avatar CK MacLeod says:

                Oops, you’re right. There are a limited number of “allowed tags” for non-registered users, by default, and, yes, it is considered sitewise unhygienic to be letting just anyone use ’em in an open text area – so, if they’re not specifically white-listed, like the ones applied by the comment-formatting buttons, they get stripped out.Report

  3. Avatar CK MacLeod says:

    Ahem – the kindly developer of Simple Comment Editing made a change, on request, and I’m uh trying it out here.Report

  4. Avatar CK MacLeod says:

    sce fixed?

    EDIT: Absitively – should be a little less obtrusive this way.Report

  5. Avatar Vikram Bath says:

    I know there are others, but this is another example of how the new commenting features have changed how conversation happens. Burt is the post author. But most of the comments are responses or responses to responses to the first comment by Damon.

    • Avatar CK MacLeod says:

      Thanks for that. I’m still looking forward to the day that we get a comment thread in multiple snakeback mode: I’ve seen several reach from snake-left (pink) to snake-right (blue), but none yet from s-l to s-r and s-l again, much less full-on multiple sl’s and sr’s. Having a breakpoint (formerly max-depth) at 10 makes that harder to achieve, but I think we have it in us.Report