 So, I guess a couple of things. The first is that the hierarchy that's shown on the left now is Bob's hierarchy. In the last version of the demo, it was the natural file system hierarchy on the server, which is not terribly interesting to anybody as I understand it. But this is Bob's hierarchy. So, if you pick, for example, interfaces, we see six columns worth of documents as we mouse over a document, the corresponding file is loaded up over on the right. The gray stripe, the vertical gray stripe indicates where you need to stay with your mouse that you're going to keep the current column expanded. If you move to the right or to the left, you'll expand the column to the right or to the left. So that hasn't changed. That's still pretty much the same. The other thing that hasn't changed mostly is that if you pick a category like essays, for example, that has a huge number of documents, we won't show them all at once. We'll actually create an outline of all of the subcategories of essays in this case. And selecting a subhead, a subcategory will bring up the corresponding categories, which again you can mouse over and bring up the pages. So nothing new there. I could do two things that are new. First of all, if you search. Now it's not a string search it actually does a Google web service search against wiki dot or excuse me code dot j software dot com. So the results you get back are a little more reasonable. There's a number of different types of search. There are actually full text search hits. And again, we've got squeezed columns compressed columns and you can move to the right to decompress a column and and see the corresponding to the corresponding pages. The way that the amount of time it takes for the web service to come back is, I think inappropriate to the application. I think it should. I'd like to have a much more responsive web search to the point where as you're typing characters were showing results. And it turns out that SQL light has what is apparently a pretty good full text search facility. So in principle, you could populate a SQL light database with the full contents of the wiki and do full text searches against it and its performance according to the documentation is pretty good on the order of 10s of milliseconds to get a response back rather than the second or more that I'm waiting that I waited to get the response back to the search for random, for example. I'd really like to do that the challenge. And I don't know how big a challenge it would be is that the SQL light binaries that with J do not include the full text search facilities so I need to swap in different binaries I don't I don't know what the implications of that are. The scope hasn't changed the one thing that has one major difference major improvement I guess is forums. So the forum interface currently is a very web style interface where you lower yourself into a cave and then you lower yourself into a deeper cave and then you find an actual forum post and then you climb out of one of the caves we love yourself in the approach I took here was to say we'll look here all the years that we've got archived 2006 through 2023 and for each year here all the months we've got archive. And as you mouse over each one will show you the subject matter, excuse me the subject headaches the subjects from the threads for that month. And I think I've been through almost all of the years and all the months and they all fit on my screen so there's no need for a scrolling mechanism. And then what happens is as you select a category excuse me a subject. You see the first message in that thread and you also see the people who contributed to that thread in order, and you can load up their contributions on the fly. And I, I think what this does is it changes the review of the forum archive from being sort of a spelunking expedition, lowering yourself into a cave complex and sort of groping in the dark on the one hand and then pulling yourself out again to lower yourself somewhere else to more like sitting on the couch leaping through a magazine. You're not highly motivated you're just sort of wondering if something interesting shows up. And I think this experience is closer to being on the couch and leaping through a magazine, then it is to what I characterize as spelunking. And I think what that could do, and I, this is just my first path I don't say that this is in any sense the right answer. I think what they could do is drive a lot more traffic to the forum archives, if we could make that work. And that's about it for new stuff. One of the things I'd really like to do is integrate, sorry, is make search work with the forum archive, and come up with some way of showing results in the context of this years by months matrix. One of the things that's interesting about this is, this is actually a four level deep hierarchy that you're navigating you're navigating years, within a year you're navigating months within a month you're navigating threads and within a thread you're navigating messages. And, but it doesn't feel like that it doesn't feel like as I say a spelunking expedition it feels like you're looking at a satellite map and just moving around on that map. Another way to think about it as you're moving ever forward, there's no such thing as going back. I believe that's all I have for this week. There are comments or questions I'd be very interested. I think of ideas I don't know how. What the payoff versus efforts can be like. What I'm thinking about here is that threads will often span months and even years sometimes. And I don't know if your UI can kind of take advantage of that. Another question is the whole, there's there's a whole whole tangent related to searching that I that I have various thoughts on I don't know. Well, please go ahead. Well, there's okay there's three different ways of approaching this is there's local searching as remote searching. And because of web view limitations I think we're your favoring local searching although if you could spin off a get a few questions a thread and then be the result into the web view that would probably let you take advantage of although with a little bit of lag because there's you still lag for remote. So that's that's one perspective another angle on this on the remote side is there's web, you know, like Google searching where you constraint to a site. You can use use a site parameter and then you say, like, and you can even constraint the forms by throwing a pipeline and it depends on whether they've indexed it that you know that that would get you that's that gives you a quick implementation even if it's not a highly performant you know really low latency presentation, then the local side, where you get your high, your, your high speed low latency responses, if you've done it right. There's, there's there's the SQLite approach they mentioned where you might have to do new binary so there's also J approach I should put my There's J approaches where is just first off there's a flat text search, like what I think Chris Burke did if you which runs into limits once you start getting into certain volume of documents but it's easy to put together you just string together documents flat thing and you do a binary search on the There's also indexing, which gets into where SQLite it probably has stuff pre built for you worried and Jay you'd have to build yourself where you say, here's words that are likely to be searched on here. Here's their locations and or here's, you know, here's a pre a pre built search for that. It's a space time trade off and and also a lot of programmer time needs to go into to make it work well that's one of the possibilities that I don't. It's probably not this stage of the game where you want to be focusing your attention. It's interesting I go ahead. I'm just winding down that's fine. It's interesting I tried to use the native wiki search facility, and it works. Additionally, the numb header, which indicates how many results you want does not seem to make it past something. So I can specify numb equals 100 for example but not equals 100 is not get passed through to the Google search service. And the result of that is no matter what I only get 10 or fewer results back. I'm actually using a Google meta search facility, which I paid the princely some of $30 a month. And that gives me that gives me my hundred. I assume that's a problem that could be fixed with wiki search either by fixing my code or fixing something else. I'd much rather use the native wiki facilities, but even more than that, it looks like SQL light does have me stemming and building inverted indexes and full text searching solved problem. There's no good reason, I think, for anybody to spend time, I'm re implementing that and SQL light, according to the documentation as I read it seems to have a very good implementation of full text search which is certainly good enough for what it is that we'd like to do. Now, whether it would handle code search, like, well, I don't know special characters is it indexing on colon, for example, couldn't you say you do a wiki search, and you do like limit equals 500 and you still get only 10. Yeah. It's entirely possible I'm doing something wrong. But when I do the same thing with the Google service, I get my I get the appropriate number of results. I forgot you mentioned that it but I if I had to guess and I haven't done any looking into it because you moved on to the Google and that solved that problem. I would guess there's a setting within a media wiki that limits that search. I'm just guessing at that because I know other areas like my trees and stuff there's limits to how there's you can set up and have a setting to whatever you want. That's interesting. Yeah, I don't know I assume it's entirely flexible I'm not not concerned about it and I don't mind using my own service, sorry the service to which I subscribe for now. I mean, Bob, you know when when you ran the startup the setup procedure. It was two minutes of downloading files and filling up tables and it was just grim. And that's nothing I think to do it right we need to download and fill up download a lot more files and fill up a lot more tables to make it work. And my current thinking, and I, I tried not to think about deployment early on but I think we're far enough along now or it's reasonable to think about it. What I'm imagining is that you would produce on the server, a single SQL light database a single SQL like file that could be downloaded by a user of this application. And there'd be a certain cost to that but the, the certainty of downloading a single large file versus downloading thousands of small files and indexing them, and so on. I think is very attractive. I'm imagining I mean I'm my ISP can handle 100 megabytes and I blick it's not it's not a big deal. I think it wouldn't, it wouldn't be a bad idea. I had just now tried doing a media wiki search, and I said a limit of 500, and my search gave me 393 results back. So, would you do me an enormous favor and send me what you passed in or send it to Bob and you can send it to me. I really appreciate it. I think I have your email address I'll send it to you and Bob both. Okay, thank you. I think you'll see Ed's email when I send out the invitation to the wiki. I don't remember if I tried it. I think maybe if I do it from a web browser it's okay but when I do it with get HTTP it's, it's a problem. The biggest difference I see is I think, Raul, you've got administrative privileges don't you. I'm doing it. I, I'm doing it from Jay, which means I'm, I'm not logged in. Okay, okay. Because let's see got some right. Yeah. So, so because I've had the benefit of actually playing with this. I've got a couple of things just to mention about the interface just things I've noticed. One, I really liked the way that sliding up and down with the blue highlighting on the on the left bar works. That's, it took me a while to figure out at first that you only you only slide when you are over the blue, when you're over the white is when you can select. It's a really nice elegant way of doing things. And I like the feedback that gives really, it's, it's great. One thing I did notice, and this might be, I'm not sure how this fits in. Last week, I think we were talking about what happens when you actually click on a category heading. You're clicked on essays right now. I don't see of those because they're my categories that's showing you a category page, but it doesn't actually show you the category page doesn't so if you go up and highlight our three at the top of there. It's going to show me art anger essays index it's not actually showing me the category page which will have content on it. I think, and it's entirely possible that the crawler is glitchy. I will not answer for that. I think that essays are three is showing the pages that are associated with the essays are three category page. That's exactly what it's doing. Yeah. But it's not showing what you're looking for. No, it's not showing the essays are three page. The essays are three page is the category pages walk category pages are never shown. They are simply a source of structure and a source of pages that are owned by each of the category nodes, but I never show the category pages themselves. That's true. I could. That's my question is if you could because those those category pages as we're developing context content for the different categories will become more useful. So where I noticed is I went down to the contributing to the wiki category, and I found I couldn't actually see the stuff I've been developing on it, because I can't actually see the actual category page. I can only see the things that's attached to it, which is useful, but all the stuff I'm putting in is in that under that green right now. Alright, so it's unfair to think of the category pages as strictly a source of structure and organization they have actual content in their own right and need to be displayed. I'm with you. I can do that. Okay, because that's that's I found that out this morning I was playing around with it again I thought. Well, the thing is the other part of it the other part of it that I might be able to do. And again this is probably me bending things to fit your interface so if your interface can adapt to it that's great. But the other thing I could do is I don't have to have these categories hidden. I've hidden them now so they don't show up on the pages. But because they don't show up on the pages I don't have the option to go over to the page. And then go down and find that category like for instance I can't go to help templates or something which is a page and scan click on that category to get back to it because on that page. The category doesn't show up it's hidden. So I've kind of. You wouldn't be able to anyway in this tool because. Oh no I guess to what. Yeah, that was the other thing I can do I can click on on your, your, your right panel just like it was a web page I can go all over the place with that. Once I'm looking at it once I've selected it I've got full control. Absolutely. Yeah. When I view source so I haven't gotten to try to go in and edit, but if I made edits there my guess is because this is actually sequestered, you know, in my in your in your file there. It's actually not going to be making changes back to the actual actually the current version is not using a cache anymore. It's actually pages on the flat, but okay, you're not logged in. No, but there's a login link. Yeah, the login links there I could log in and make changes. Yeah, yeah, if it's if it's live as a wiki page absolutely I could do that. Yeah, which is that's cool. It gives you another way. It's the old school way of doing things but it does give you another way of exploring things. The question is, does how does WebView support cookie tracking which is necessary for login. And also when you mix metaphors when you when you you've done stuff on the left and you go over here and you and you start navigating to other pages. Does that still preserve the original context the original cookies. Yeah, I don't have an intuition about that. And one thing you've done. I'm not seeing it on this version but when you select, you were actually showing the URL above the middle pain is that right is that still there. Oh, yeah, if you click on, yeah, and I put this in for your convenience although maybe it doesn't really matter if you can if you can edit the page directly. I couldn't figure out how to get to the paste buffer but you could command seed that URL and paste it into a conventional browser and do what you do. That, that text element is just a jqt text element right. Yeah, exactly. Yeah, there you can you have control you have events you can actually pull up the contents of that when you do certain things so you just have to put a button that says I want to use that, and it will actually give you the contents. Like you know I'm not sure I understand that. Yeah, the question is, can I put it in the paste buffer for the Mac page buffer. That's the part I'm clear on. I think wd clip paste though is still supported I should double check that. Oh yeah. I believe it is. Yeah, I'll look it. Yeah. I wonder what'll happen if I search for that. It's a single run together word. Yep, looks like the current command reference still mentions paste. It's two words run together as a command. What's happening with my video I'm not on my usual machine so I'm going to keep it off. If you can see it looks pretty weird. That's sideways sideways. And started. Yeah, yeah. So I think I need to reboot things so I won't do that now. Yeah, no don't don't because we don't want to lose you. We can hear you fine. Yeah, I do have some. I do have some things I've been doing, putting categories on NYC jug pages, which show when there's a chance I'd like to show. Oh yeah, by all means, let me stop my share. Well actually I wanted to show one more thing. Oh, if you can grab it back and I just want to show one one more thing because I said you might have a way around it, but sure. And it's to do with the with the the search of the forums. When you go to the programming forum. So you go to 2007 if you go over the names there it's going to jump right now it stays in 2007. Yeah, okay. Because when I was doing it it was jumping around on the numbers as I went up and down but I guess I wasn't getting far. Yeah, that's possible. And it's also possible that it's just a bug. So thank you. Not. Yeah, no it's locking just the way I would have thought just like what I was finding was when I say I set 2008 and I want to go to September, when I went over if I if I came up, I was probably too close to the greens I'm guessing. Because you can go over now set September right. Yeah, no that's right that's exactly how you'd want it to work. Yeah, I don't I'll check that again. But that's the only. Yeah, could you and if you could, if you if you're feeling really ambitious if you could make a video of this behaves that would be great. Yeah, yeah, I will do that. I've learned over the years, only to respect bug reports, even if they don't seem reasonable right at the outset. Okay.