[X] My Assistant

The max number of results per page is now 100. Paging Enlargement III was removed and will be refunded Soon™

I did not receive a refund (IMG:[invalid] style_emoticons/default/sad.gif)

I did not receive a refund (IMG:[invalid] style_emoticons/default/sad.gif)

No one did, yet. You will, Soon™.

The new search update is amazing. Thank you to everyone who helped in its continued development.

The current ESR is new enough that it shouldn't have been an issue. The previous ESR would have been too old, and some Firefox derivatives (Waterfox, PaleMoon, possibly others) are also based on an older version of Firefox that does not support the operator in question.

Thank you, from a Seamonkey user that may have possibly been impacted. Not that I use the bar.

The site cannot support old browsers. You will need to update if you have a problem. You cannot keep using the same old browser forever.

I can, and I do.

This post has been edited by dragontamer8740: Dec 28 2022, 07:21

Let me know if you encounter any search queries where the settled estimate is *vastly* inaccurate - that is, off by more than 50% or so.

Searching "focus anal" -anal$ says "Found about 5,611 results" but in reality there are about 50 (~1% of estimate).

Searching "focus anal" -anal$ says "Found about 5,611 results" but in reality there are about 50 (~1% of estimate).

It's not wrong, it just can't estimate exclusions, so the estimate is before results are excluded.

Ah okay, I thought it might be something like that. Sorry if the thing about exclusions was explained before, I just saw something about it under- rather than over-estimating the result count, so I was surprised by this search. This makes me wonder though if the estimate for exclusion searches should be handled differently, because having this bogus estimate will be confusing to the standard user.

And for simple searches with a single exclusion you could try estimate("tagA tagB -tagC") = estimate("tagA tagB") - estimate("tagA tagB tagC") I suppose, if that is somewhat practical.

Ah okay, I thought it might be something like that. Sorry if the thing about exclusions was explained before, I just saw something about it under- rather than over-estimating the result count, so I was surprised by this search. This makes me wonder though if the estimate for exclusion searches should be handled differently, because having this bogus estimate will be confusing to the standard user

At some point it was showing both filtered and excluded counts (i.e. "You filtered x results and excluded y results from this page), but it was a) kinda cluttered, and b) not entirely consistent since *some* exclusions are done earlier, so I dropped that. It's not feasible to provide an estimate that takes exclusions into account in general.

Right now it's not entirely consistent since if the results fit on one page it shows the count *after* exclusions, and I'm considering changing that just to make it consistent if nothing else.

While I read the entirety of your post justifying the search feature, I'm not sure if "it'll be 3 seconds now, but it'll be 6 seconds in two years" is entirely usable as a response since we're just end users and we obviously don't have access to the kinds of things you do to allow us to see that projection. You're the admin and developer so of course you probably know better, but if you're saying your site will grow exponentially by a factor enough to require that kind of workload then man, sounds like the normies are finally finding us after over a decade of you in service, huh? I have noticed a direct uptick of foreign nationals as well.

I actually would've been fine with you giving #2 - i.e. something like a "more results" button that allows the server to do the search in parts, and if the server detects that the user is actually going through the pages manually it would cache parts of it for the user - i.e. truncate the search and you just hit a button and it'll give you the rest of the paginated numbers I don't know. Sounds like you didn't want to do the idea anyway.

I know you felt the search change didn't warrant a "PTR/beta test" server but you should well know that users HATE UIX changes. Especially ones that cause us to require more time/effort to do the same thing. You use Youtube, you should know Google is notorious for pulling stunts like that - their current Android client is absolute trash with numerous ridiculous UIX changes; same for the website.

With Youtube they can obviously do A/B testing (but they've never seemed to have gotten the clue people don't want unnecessary UIX changes) but I guess for you, you don't have that option and you have to do a separate test server or something, but yeah, it's easy to forget what users will think when you make changes. You've made many other changes before and you've generally hit them all out of the park where I never felt the need to post on the forums about it. But this change was a pretty big one and it's essentially a core feature change so to me it makes complete sense that you'd have a bunch of people complaining about it.

I'm sure we'll adapt. I'm in the "I don't like it" crowd but if you are saying you felt like you had to do it well, you had to do it then. Thanks for giving a better explanation with your followup post.

This post has been edited by Keirnoth: Dec 31 2022, 12:09

I know you felt the search change didn't warrant a "PTR/beta test" server but you should well know that users HATE UIX changes. Especially ones that cause us to require more time/effort to do the same thing. You use Youtube, you should know Google is notorious for pulling stunts like that - their current Android client is absolute trash with numerous ridiculous UIX changes; same for the website.

With Youtube they can obviously do A/B testing (but they've never seemed to have gotten the clue people don't want unnecessary UIX changes) but I guess for you, you don't have that option and you have to do a separate test server or something, but yeah, it's easy to forget what users will think when you make changes. You've made many other changes before and you've generally hit them all out of the park where I never felt the need to post on the forums about it. But this change was a pretty big one and it's essentially a core feature change so to me it makes complete sense that you'd have a bunch of people complaining about it.

I'm sure we'll adapt. I'm in the "I don't like it" crowd but if you are saying you felt like you had to do it well, you had to do it then. Thanks for giving a better explanation with your followup post.

Users don't like change for like 5 mintues, get used to it and then are like yeah this is amazing. People are just usually temporarily upset, and you just to ignore them until they get used to it.

Regardless of how much testing was done, the changes after the fact that allowed for where it is now wouldn't have really come to be until the same stage anyway. So it's kinda pointless to suggest it would have had any improvement in the situation. People would have still be upset but maybe like 1 less person would have been as upset.

And as far as options go for the new search, again this was the only realistic solution. So it's just what it is.

This post has been edited by Cipher-kun: Jan 1 2023, 07:12

Not sure if it's expected, but the result count seems to fall back to a lower number even after reaching a more precise one.
I searched for "female:pantyhose$" in "Game CG" about a week ago, it said found 1000+ results but the actual number was 3000+.
The result count hit ~2200 after I reached the last page, but it's back to 1000+ now.

Not sure if it's expected, but the result count seems to fall back to a lower number even after reaching a more precise one.
I searched for "female:pantyhose$" in "Game CG" about a week ago, it said found 1000+ results but the actual number was 3000+.
The result count hit ~2200 after I reached the last page, but it's back to 1000+ now.

The search cache only has that much memory to work with, so if it's been a while, the record has probably expired or been punted from the cache. I'm planning on splitting the memory cache into a separate one for the search engine stuff, but that won't be until I build a new web cluster.

Users don't like change for like 5 mintues, get used to it and then are like yeah this is amazing. People are just usually temporarily upset, and you just to ignore them until they get used to it.

Regardless of how much testing was done, the changes after the fact that allowed for where it is now wouldn't have really come to be until the same stage anyway. So it's kinda pointless to suggest it would have had any improvement in the situation. People would have still be upset but maybe like 1 less person would have been as upset.

Save the smug attitude for trolling reddit, that's dead wrong and shows a gross lack of comprehension of why this thread was such a shit show. The site was temporarily unusable except in a very specific use case of front page browsing. People don't just get used to that. A test instance would have seen immediate feedback and could have avoided the user base meltdown, not that I fault this setup at all for not having an easily accessed and permissioned test environment because that shit is a lot of work. Besides, people weren't reacting to simple feature changes, they were reacting to the loss of ability to navigate the site. Maybe you're right though, maybe all software companies should jettison their testing departments and merge all their code straight to production. Testing is pointless, it never results in any improvement and it's pointless to suggest testing improves software. The whole QA industry is just a bunch of grifters.

Anyway, switching gears positively for what I came here for, I wanted to say I personally love the new way to navigate "pages" by clicking the circles and the way it lets you know where you are in a search. I love it way more than the old way and way more than any other manga site navigation out there, and I think this final version of a slider solves that user pain point regarding browsing brilliantly. The datepicker navigation is clunky, but gets the job done (and I'm not the target use case for it anyway). And I love the new go to gallery button. Beats the hell out of having to search the particular image I'm on to then get to the gallery to search similar tags. All in all, bumpy ride but a massively improved front end experience to go with the new search engine (elasticsearch?). Kudos on being able to take all the barbed feedback and general blowback in this thread, listening to and addressing the complaints, and then building a better site out of it all to solve those complaints. Must have been hell, but I appreciate your work!

The range indicator doesn't seem to support weak tags (I guess because they have separated index?).
Should it be hidden when the query contains weak tags?

The range indicator doesn't seem to support weak tags (I guess because they have separated index?).
Should it be hidden when the query contains weak tags?

Yeah, that's definitely acting glitchy. It should be able to just use the rangemap for strong tags as a pretty good approximation, but I'll look into why it's not.

Save the smug attitude for trolling reddit, that's dead wrong and shows a gross lack of comprehension of why this thread was such a shit show. The site was temporarily unusable except in a very specific use case of front page browsing. People don't just get used to that. A test instance would have seen immediate feedback and could have avoided the user base meltdown, not that I fault this setup at all for not having an easily accessed and permissioned test environment because that shit is a lot of work. Besides, people weren't reacting to simple feature changes, they were reacting to the loss of ability to navigate the site. Maybe you're right though, maybe all software companies should jettison their testing departments and merge all their code straight to production. Testing is pointless, it never results in any improvement and it's pointless to suggest testing improves software. The whole QA industry is just a bunch of grifters.

That's an exaggeration. It was not unusable; it was simply unusable to you.
I will say that prior to having a 'goto' box or date seeking (see my signature), it was quite annoying to do so, but it was not impossible to type numbers into URL's or do math.
The negative feedback was still deserved. But the really stupid thing about it was everyone ignoring the reasoning for the change and asking for page numbers to come back.

Also, you may not realize this, but the deployment was in fact staggered. I was seeing it before it was pushed site-wide.

Given the state of many websites and programs these days, it is clear to me that testing works for stability purposes but not for making a usable, performant, quality product. Otherwise electron apps would not be so prevalent. I think companies already essentially ignore QA and make paying end-users their beta testers.

This change was made for performance, and that is an honorable goal compared to the things so many prioritize.

I appreciated the transparency and technical justification for the change, which is also so often lacking in QA.

Also I think people have some right to be smug, because the reaction really was ridiculous. I was more upset about the low level arenas being hidden from high level users than this, because this change had a really good explanation/justification.

Anyway, switching gears positively for what I came here for, I wanted to say I personally love the new way to navigate "pages" by clicking the circles and the way it lets you know where you are in a search. I love it way more than the old way and way more than any other manga site navigation out there, and I think this final version of a slider solves that user pain point regarding browsing brilliantly. The datepicker navigation is clunky, but gets the job done (and I'm not the target use case for it anyway). And I love the new go to gallery button. Beats the hell out of having to search the particular image I'm on to then get to the gallery to search similar tags. All in all, bumpy ride but a massively improved front end experience to go with the new search engine (elasticsearch?). Kudos on being able to take all the barbed feedback and general blowback in this thread, listening to and addressing the complaints, and then building a better site out of it all to solve those complaints. Must have been hell, but I appreciate your work!

I mean, the main 'hell' bit would be the programming. People getting uppity about things they don't read about or understand on a free site isn't something I'd work myself up about.

I only wish the jump text entry field was visible by default (or else was invisible but actually existant in the DOM so i could use custom CSS to fix it) so I didn't have to enable javascript. But I can cope and I'm very likely not an average user.

This post has been edited by dragontamer8740: Jan 6 2023, 21:41

Save the smug attitude for trolling reddit, that's dead wrong and shows a gross lack of comprehension of why this thread was such a shit show.

Call me smug all you want, it doesn't change the fact that I've read through every post in this thread. I think I understand.
People were upset at things they didn't understand. People refused to read before posting. People thought pages were fundemental and everyone used them and were obviously wrong. That is why it was a shit show.

Inbetween all that was a very small group of people providing actual feedback and opinions on it. As well as ideas to address the problems and add improvments. Some of which (the new range navigator) required the rollout of the new search to be finished.

Change was made for a foundational reason and that was ignored because the UI changed. Was it perfect at the start? No. Was it ever going to be? No.

The ride was bumpy but was it ever going to be worth changing the entire rollout when the development team size is 1? No.
People have come to expect that major sites have VC levels of money and unlimited amount of developers but that's not always the case.

new search engine (elasticsearch?).

Like the old one I believe it's completly custom.

This post has been edited by Cipher-kun: Jan 7 2023, 06:31

I've read through every post in this thread.

Oof.

The range indicator doesn't seem to support weak tags (I guess because they have separated index?).
Should it be hidden when the query contains weak tags?

This should be fixed now, let me know if you still see any issues with this.

Search 'c88 bsn' returned nothing
But it should return something like '(C88) [Mercuro (Various)] ♂ Kemo Lingerie Book+ [English] [BSN]'

2023-01-12 fixed

This post has been edited by kxmp: Jan 12 2023, 17:16