So, it's finally release time!

http://wakaba.c3.cx/releases/wakaba_3.0.0.zip
http://wakaba.c3.cx/releases/kareha_3.0.0.zip

I decided to bump the version number up to 3.0.0, partly because of a number of new features, and partly because there's been lots of messing around in the guts of the scripts, which means there are probably some new and interesting bugs. I do not recommend installing these scripts on any busy boards without doing some testing first, or waiting for others to test them for you. Conversely, testing is very welcome. Report those bugs!

Specific information will follow in the next posts.

>>99
Not in Kareha's case. Once threads/replies are created, the only way to keep them up to date is to modify the class names by hand in each html file.

http://pyc.h8r.net still exhibits the >>93 bug, even after updating wakautils and rebuilding the caches. Does the fix affect only new posts?

>>102

What in the... The fix has disappeared from all my files, too!

All right, I added it back one more time, and uploaded new files. Didn't bother to bump the version number, just re-download. The only file that's affected is wakautils.pl

>>103
Snacks is covertly undermining your efforts.

Also, new bug found: http://wakaba.c3.cx/soc/kareha.pl/1098492851
It happens when you put more than 3 whitespaces in front of a line of text.

Not a bug, it triggers code mode. Four space or one tab will make everything <pre> formatted. It's copied from Markdown, but it's a bit unfortunate since it's sort of easy to trigger by mistake. Haven't come up with a better solution though. Using only tabs would be hard to trigger accidentially, but most web browsers make it impossible to input tabs by anything except copy-paste.

http://validator.w3.org/check?uri=http%3A%2F%2Fwakaba.c3.cx%2Fsup&charset=%28detect+automatically%29&doctype=Inline

is this a bug:

http://4-ch.net/iaa/kareha.pl/1132249274/

>>108

Eh... Let's say it isn't.

This is the ■ ▲ ▼ for the 6th thread on the /dis/ board, and it's located at the very top of the page. css bug?

It's worth noting Pseud0ch is still borked in Opera. Namely, the [del] links. I'm not certain whether it's the -1.3em or the relative positioning, but deletebutton should be replaced with one of the ones in the other CSS.

BTW, Amber seems to work fine with 3.0, at least for normal posting modes.

>>111

That's not 3.0.

>>112

No, it doesn't. It's missing a ton of frames.

See here: http://forum.akatsukimanga.com/index.html

It's missing a ton of frames? So what? The differences are trivial. It also works better than Pseud0ch (cough).

http://wakaba.c3.cx/releases/wakaba_3.0.4.zip
http://wakaba.c3.cx/releases/kareha_3.0.9.zip

Latest batch of bugfixes:

• Should fixes the validation errors in >>106.
• Fixes a warning in Wakaba that was spamming logs.
• Adds the missing link for deleting threads in admin.pl.
• admin.pl now never uses XHTML mode.
• Fixes a big where spaces in filenames got encoded as + instead of %20
• Now strips out numerical entities bigger than 1114111, which apparently causes Mozialla YSODs.
• Pseud0ch (and some other styles) fixed to work better in IE. I had no idea it wasn't working, you people need to tell me this stuff.

>>111 a missing position: relative; for sagethread in pseud0ch.css. This bug ist still present in kareha 3.0.9.

>>116

Thanks! Fixed and re-uploaded 3.0.9 without bumping the version number.

Some of the "Return" links match the referring url (eg: the error page when a thread doesn't exist) - but when there isn't a referer, those links go nowhere.

Did you make any substantial changes to how build_cache() and build_cache_page() work? The changes I did for my board were done with a nebulous understanding of the inner workings, so...

Umm... In Wakaba, no, in Kareha, yes. As I recall.

Pretty much. Not planning any big work in the near future.

>>122
So we will not have a working post preview in the near future?

>>123
Post preview shouldn't be implemented in the core functionality. It would work best as a browser extension.

>>123

It's not a very big thing, actually, if I implement it with XMLHttpRequest as mentioned elsewhere.

>>125
Then please do it so. :)

Kareha (in Futaba mode, at least) seems to have a bug where clicking a post number on a main page (to reference the post in a reply) produces a something like http://dongs.hadoken.net/kareha.pl/1134468901/#i%3E%3E1 in the URL bar. The link actually works for inserting post references -- that's why I'm not sure if it's a bug or not.

What happened to the dumplog.pl file came with kareha 2.x? How do I decode the IP address of a poster in kareha 3.x?

Aah, I didn't leave in any automatic way to do it, but you can edit admin.pl and change <var $masked_ip> to <var $ip> to have it show the real IPs instead of the masked ones.

age for >>127

>>127 isn't a bug.

http://dongs.hadoken.net/kareha.pl/1134468901/#ixbox%20is%20hueg
heh

i've found a bug in kareha... if i set CAPPED_TRIPS and rebuild caches, all posts on the board disappear from index.html and the thread views, but are still in the files in res/. if i comment out the line for CAPPED_TRIPS, the posts reappear when i rebuild caches.

What did you set it to?

>>134

use constant CAPPED_TRIPS => ('!hoTarufiRE!!yUegcdbn'=>'<span title="admin">!hoTarufiRE</span>');

That's very weird... I can't think of any way that could happen in the first place. I'm even using CAPPED_TRIPS right now on this board and it works fine.

Is there a function yet that limits the number of times you can post / start threads in a given time?

That can be controlled to some extent with the ancient RENZOKU options that were inherited from Futaba. For Kareha, no such options exist.

Maybe there should be. Just to be able to prevent people from scriptspamming.

Also: Is it possible to define a keyword that one could use as name or link and that would sage a thread but reveal the ID code (for boards that have it usually turned off or turned off when saged)?

>>140
yes, if you don't mind writing a little bit of perl.
i would do it for you, but i'm too lazy.
maybe WAHa will do it for you, but he's almost as lazy as me, so i wouldn't count on it.

http://wakaba.c3.cx/soc/kareha.pl/1100499906/q81 gives

Software error:

Undefined subroutine &main::get_reply_text called at /home/waha/public_html/soc/kareha.pl line 875.

For help, please send mail to the webmaster ([email protected]), giving this error message and the time and date of the error.

>>142
>>72

So, will Kareha maybe get its needed banning script soon maybe? Please?

Don't you think that Kareha has awfully bloated and complicated code for what it does? Come on, you can write it with maybe two hundred lines, and you don't have to include the sources of a http server... :)

The idea is very cool, but I'd be afraid to run this code on my server.

> you can write it with maybe two hundred lines,

then why don't you?

>>144
Use Apache/IIS/whatever. That kind of functionality is really not needed in Kareha.

>>145
What kind of bloat and complication are you referring to?

Actually, I might end up adding some banning functionality, just so I can put in traps to automatically ban spammers. But we'll see.

And >>145, are you sure you know "what it does"?

I think >>145 is looking for something more like http://www.graffitiwall.net/

>>146
I'm going to ;)
Got something like >>149 so far...

>>147
You know, including the sources of perl libraries you use just to make some changes in them... Copy-paste programming...

I know, I know, it's Perl, it must be ugly ;)

>>148
I thought so at the beginning, but the more I read the sources the less sure I am ;)

>>149
Much simplier, indeed, but lacks some of the less-frequently used functionalities of this board :).

Sorry for the harsh words, but I really think it could benefit greatly from some code cleanup.

> You know, including the sources of perl libraries you use just to make some changes in them...

Pretty much the only external library that is included is MIME::Base64. This is because it's not in all default installs, and most people can't install external libraries. A lot of other code is written from scratch, but duplicates functionality that exist in third-party modules, but since lots of people can't install those, I can't use them if I want people to be able to use the script.

>>151
If you want to be free of dependencies, why don't you make it into an operating system?

>>152

I always hate programs that require root access to install. It's a huge damper on popularity.

Ever written any software for public consumption, >>152? --;

>>152

To give a straigh answer to a rhethorical question: Because if people can't install modules because they're on a shared host, they can't install a new operating system either.

I'm not avoiding dependencies for the sake of it, I'm doing what is needed in order that the largest number of people be able to use the software.

Well, besides switching to PHP. There are limits.

I don't think using PHP would gain you much. Very few hosts that offer PHP don't offer the other two P's.

The main thing it would gain is that idiots who limit Perl to running in /cgi-bin/ usually don't apply the same restriction to PHP (making the restriction pretty useless in the first place, needless to say).

What's wrong with putting all the scripts in cgi-bin/ and your custom perl modules in some directory in your home?

It just works, unless you try to use some scripts with hardcoded paths in them.

It's not rocket science.

Most Perl modules use compiled code. Even for those modules that are pure Perl, most people don't know how to do this anyway. It is more work when installing, and people are more likely to not bother.

As it is now, you upload a handful of files, chmod, and it works pretty much just like that. The fact that your program will be more popular the less trouble it makes for your users shouldn't be this difficult to grasp.

You mean there are hosts that have all those fancy image processing libraries you use, but lack base64 modules? Come on...

Ban-script.

>>160
Wakaba can work without any of those libraries. It's not a good idea, but it's possible.

> As it is now, you upload a handful of files, chmod, and it works pretty much just like that.

I think it's quite telling that there are many people who have difficulty doing only this. If anything, the installation should be made significantly more easy.

So if they have problems with editing config.pl and using chmod, does anyone think they'll be able to do it if they need to install additional perl modules too?

The point is to have a running board. Making users jump hoops isn't.

>>161

Post preview.

I see, so ugly code and sloppy programming is for dummies...
Sounds fair. Sorry for my bitching, didn't think about the "marketing" part of programming.

Forsaking code reuse through external dependencies might not appeal to your sense of aesthetics, but I will take offense at "ugly" or "sloppy". If you want to make claims like that, you'd better come up with a better reason than not using external dependencies.

Well, wakaba (at least) could be cleaner. It's a prettier than the alternatives, but it has certainly grown a few hairs over time.

Yet I think worrying about this is silly. There's one main developer. This individual has a limited amount of time. What would be a better idea in spending it: refactoring code that isn't broken, or fixing code that is? What about adding requested features?

From a user's point of view, the former gives no return. Remember, successful software is "good enough". The perfectionists never go anywhere (except in circles), while the imperfect software is actually being useful to the market.

This is reality.

Well, right, for Wakaba, what I'd really like to do is cut certain parts up in modules, but there are only so many hours in the day. The moderation interface in particular could really use some sort of plugin system, so one could implement various kinds of other moderation options and features.

Anyone feel like doing that, go ahead.

>>166
Sorry, didn't mean to offend you, just got a little carried away.

I'll just shut up and go now.

Btw. Somehow didn't say it before, but it's really the best kind of "forum" I ever tried, and the idea that the code doesn't look as beautiful as it could comes from the assumption that it's really top quality.

It happens to the best of us. Donmai, donmai.

My host has Perl v5.6.1 and I'm having trouble installing kareha 3.0.9. I followed the instructions in the documentation but I get a software error when try to run kareha.pl

It said

Invalid [] range "}-\x" before HERE mark in regex m/([\x{100}-\x << HERE {ffff}])/
Compilation failed in require at templates.pl line 3.
BEGIN failed--compilation aborted at templates.pl line 3.
Compilation failed in require at /kunden/homepages/5/d92823313/htdocs/bbs2/kareha.pl line 14.
BEGIN failed--compilation aborted at /kunden/homepages/5/d92823313/htdocs/bbs2/kareha.pl line 14.

I also tried to install an earlier version (2.0.4) in another directory and that works just fine. Anyone know how to get 3.0.9 to work correctly? Any help would be greatly appreciated.

Oh, I seem to have accidentially gotten a 5.8-only feature in there without any checking for earlier versions. I'm pretty sure that line is not important, so you can just open up wakautils.pl, find the line, and comment it out. I'll have to fix that.

Bug: You can make empty posts by selecting HTML formatting and typing a post that consists only of an HTML tag that isn't allowed. Apparently the tag gets removed, thus leaving the post empty but bypassing the empty post check. For instance, entering just <!-- hello --> will make an empty post, if you select HTML formatting.

I'd hardly call that a bug. There are any number of ways of making empty posts, with any formatting mode. It's hardly a problem. The check is only there to catch you if you click Reply by mistake.

<!-- hello --> testing anonymous's claims.

>>174
What other possible ways are there to make an empty post without formatting?

__
**

>>176

There are a whole bunch of unicode characters that are completely invisible.

is this a bug?
http:javascript:alert(document.cookie)

> The requested URL /sup/javascript:alert(document.cookie was not found on this server.

It might work in IE (if you fixed the missing end parenthesis), but properly filtering out all the idiotic ways that IE will go out of its way to execute Javascript is pretty hard.

i was just reading >>157 and i thought of something... if you use a host that restricts perl to /cgi-bin/, can you just make a php script that just does passthru("./kareha.pl")?

also, maybe you could implement this?
http://beta.4-ch.net/dis/kareha.pl/1130990333/

Links made with html formatting don't have rel=nofollow added. You might want to fix that before spammers start using it.

>>181

Yes, I know, I've fixed that already but I haven't had the time to put together a full release with the rest of the needed fixes.

>>182 in those setups, php is usually also restricted and it cannot run external programs. Usually -- YMMV.

The auto linker seems to not work right with urls that have a ) in them.
Example: http://en.wikipedia.org/wiki/Aspect_(linguistics)
So you have to use http://en.wikipedia.org/wiki/Aspect_(linguistics%29 or HTML formatting to link them properly.

That's a feature, because if it wasn't for Wikipedia, nobody would ever end a link with a ), but lots of people would say something (and include an URL: http://wakaba.c3.cx).

It would be a hack but you could include it in the link if the url is from wikipedia and it has a ( in it that is unmatched.

>>186 Likely not worth the work - click the link and add your own ) at the end, ya pansy!

>>187
or just replace the ) with %29

A better solution would be to include the ) if there's a matching ( in the URL, but I'm really not sure I feel like writing that regexp. It's pretty horrendous already.

If somebody writes it for me, I'll put it in, though.

#https?://[^\s]+(?<!\([^\s]+)\)# doesn't match if there is a ( in the url
#https?://[^\s]+(?<=\([^\s]+)\)# does match if there is a ( in the url
#https?://[^\s]+\([^\s\)]+\)# if you don't want to be fancy about it, matches any url with a ( in it and a ) at the end

When are you ever going to have more than one set of parens in a url? I figured never so i didn't bother thinking about it.

So for the whole thing you end up with something like
#(https?://[^\s](?(?<=//[^s]\()[^\s]|[^\s\)])#<a href='\1' rel='nofollow'>#

Er, the last one should be

#(https?://[^\s]+(?(?<=//[^s]+\()[^\s]*?|[^\s\)]*?))#<a href='\1' rel='nofollow'>#

.. fucking wakabamark

There's a lot more to that regexp currently than what yours does, though. There are a number of of characters that are ignored at the end, for instance.

Currently it is:

($protocol_re:[^\s<>"]*?)((?:\s|<|>|"|\.|\)|\]|!|\?|,|,|&quot;)*(?:[\s<>"]|$))

...except that one of the commas is a numerical entity, which can not be posted. Sigh.

One is & # 44 ; Why didn't you do (?:[\s<>".)!?,\]]|&quot;|& #44;)
It is kind of amazing how much I screwed up the ones last night, never drink and regexp =(... varible length lookbehind is perl6 even. I'll try to rewrite it with lookahead and factor in all of the current crap if this hangover doesn't kill me.

i wrote a recursive regexp that behaves exactly like the existing one in wakabamark (i think) except for including balanced parentheses. of course it's perl 5.8+ only, but since i have the regex in a slightly more readable form it might be a better place to start for doing the 'one or no pairs of parentheses' regex.

http://rafb.net/paste/results/BohIqW25.html

Suggestion: How about making a JavaScript box pop up when you ban someone in the Wakaba Management Panel, prompting to enter a comment, so that one doesn't have to copypaste the IP address into the ban Ban page every time?

>>198

Not a bad idea, probably.