All picture\post show up blank with a question mark in a blue box? How can i fix this?

DarthMalak said:
All picture\post show up blank with a question mark in a blue box? How can i fix this?

What browser/OS are you using? Does it behave differently if you load the page in a private/incognito window?

parasprite said:
From personal experience: On any website, always, always always move your text walls to a new wordpad/notes/textedit/Word/etc.

Usually what I do, but I was stupid enough not to do it and simply put, got mad (mostly at myself), just never hit me that an AJAX request would redirect me to an error page, and unlucky me got another error when hitting back, yay :P

parasprite said: (re CloudFlare blocks)
What you can do: If you get a blocked page... Send this to either me or Dari...

Nice, I'll do that next time. (Not that I know how to get in contact with Dari, unless spamming management at dragonfru.it, anyway you seem to be a go-to kind of person :D.)

On a similar note: I got a wall of text which always gives a 502 (so in frustration I made my own dtext parser, but that's another story). However if I remember correctly that used to bring me to a CloudFlare page, now it says nothing, or (rarely) the error shows up in the preview window (which is a thousand times better, even if not good):

<center><h1>502 Bad Gateway</h1></center>
<hr><center>cloudflare-nginx</center>

Maybe someone has made a "fix" already? Or it might simply be a load issue.

parasprite said: (re images not loading)
What browser/OS are you using? Does it behave differently if you load the page in a private/incognito window?

Had a few problems with that as well recently, for me it seemed to be ads.dragonfru.it/www/delivery/ajs.php not responding in a timely fashion. Might be related. (Chrome/OS X 10.9.5)

Bug: Post update occasionally (once every 200th?) fails due to deadlock (I think similar cases has been up on the table before, e.g. forum #132253)
Expected behavior: Update goes through
Actual behavior: 5xx error (PostgreSQL), Pastebin HTML
Steps to duplicate: Um... Get (un)lucky?

Chessax said:

Nice, I'll do that next time. (Not that I know how to get in contact with Dari, unless spamming management at dragonfru.it, anyway you seem to be a go-to kind of person :D.)

Oh, right. Yeah then me probably. Even if it isn't me I'll make sure it gets to the right place. '.')b

Usually what I do, but I was stupid enough not to do it and simply put, got mad (mostly at myself), just never hit me that an AJAX request would redirect me to an error page, and unlucky me got another error when hitting back, yay :P

On a similar note: I got a wall of text which always gives a 502 (so in frustration I made my own dtext parser, but that's another story). However if I remember correctly that used to bring me to a CloudFlare page, now it says nothing, or (rarely) the error shows up in the preview window (which is a thousand times better, even if not good):

<center><h1>502 Bad Gateway</h1></center>
<hr><center>cloudflare-nginx</center>

Maybe someone has made a "fix" already? Or it might simply be a load issue.

Partially. The settings for Cloudflare were tweaked in the last day or so to not overreact as much (as in it won't think a random comment critiquing furry porn is a potential SQL injection) as well as fix the problems we've been having with some uploads (particularly flash). The good news is this means the "blocked" pages shouldn't be popping up anymore during what is otherwise normal site behavior. If they do please let us know.

The preview button problems are actually a completely different issue, and are probably unrelated to Cloudflare. So far I've managed to gather the following information on this:

• It seems to only affect forum posts (at least that's what I've found from my own tests).
• You can trigger a 520 website is offline redirect-hijack with by using the preview button on a post composed of ~7930-7940 or more characters (it seems to differ slightly by browser for whatever reason). I've triggered them consistently with sequences of random letters, but it can happen with posts that aren't as well, just not as predictably.
• If you type a small comment and hit preview, then put in a long comment with markup (probably over ~7930 characters at least) and hit preview, it seems to give 500 Bad Gateway error in the preview box instead (without the annoying redirect).
• If you type a small comment and hit preview, then put in an extremely long comment with markup (mine was 45000 characters) it will actually give a 400 Bad Request - Request Header Or Cookie Too Large error in the preview box (without redirect).
• I haven't managed to reliably produce a 502 error from the preview button, but I suspect it would happen under similar conditions to the 500 error. I'd be interested to see the text if you still happen to have it.
• Tony said a potential fix might be to send it as POST instead of GET. I'm not that familiar with how HTTP requests are sent myself (and I don't know if you are), but I mention it just in case you might find it informative.

Had a few problems with that as well recently, for me it seemed to be ads.dragonfru.it/www/delivery/ajs.php not responding in a timely fashion. Might be related. (Chrome/OS X 10.9.5)

Are you still having issues with this? I haven't been on as much as usual due to finals, but I'm not seeing this behavior right now. I do remember when the ad server was down a while back you could "fix" the page loading issues temporarily by enabling adblock. Might be worth trying (temporarily of course ;) ) to see if it helps.

Screenshot of the problem happening: http://i.imgur.com/JI4yqLz.png
Link to the page shown in the screenshot: https://e621.net/post/show/652077

I can see the image now after taking the time to type this message and then going back to that page and refreshing, but when I first tried to view the picture it wouldn't show up. The example image happens to be a really recent one that just happened to have this problem with it, but MUCH older images do this too sometimes. I'm not sure what causes it, and the only thing I can do is look to the source of the image. If it doesn't have a linked source, or if it does but it's not on Fur Affinity, then I'm pretty much forced to click the back button and forget about trying to view the image.

Most of the time refreshing doesn't help, usually unless I wait several minutes before attempting to refresh. But more often than not, it seems refreshing never works no matter how long I try to patiently wait.

Clearing the whole cache and attempting to view the page again has helped sometimes, but it's not a totally reliable solution and it's annoying to have to sign into everything again.

Using features such as reverse Google search doesn't seem to be very reliable in my experience. I never find the image depicted in the thumbnail I clicked on...

Clicking resize image or download has never helped.

If it helps, on my computer I use Internet Explorer 11 on Windows 8.1, and my mobile device is a Window Phone 8.1 with Internet Explorer.

parasprite said:
The good news is this means the "blocked" pages shouldn't be popping up anymore during what is otherwise normal site behavior. If they do please let us know.

:D and I will let you know just below, hehe.

parasprite said:
The preview button problems are actually a completely different issue, and are probably unrelated to Cloudflare. So far I've managed to gather the following information on this:
...

Wouldn't call them completely unrelated as CloudFlare performance does affects previewing.

If you ask me GET is definitely one of the main problems here, GET request are usually, as far as I know, limited to 8 KiB (or even less) before web servers start acting up in one way or another (e.g. throwing errors, truncating etc). I tend to think like this: Use GET when you want to get something, and POST when you want to post something, maybe not the proper approach but that way you rarely run into those limits and you're sort of "posting" a preview, it's just not saved. I fooled around with some single byte characters (only a bunch of "a"s) and found the following (to support your claims):

Note: Any numbers below will most likely differ slightly between browsers/OSs.

• 7909 characters => 200 (ok) (for a total request URL of 8193 characters)
• 7910 characters => CloudFlare 520 (e.g. Ray ID: 1e3e1e63496a05a3 • 2015-05-09 14:33:18 UTC).
• 13549 characters => CloudFlare 520 (unknown error)
• 13550 characters => HTTP 502 (bad gateway)
• 32443 characters => HTTP 502 (bad gateway)
• 32444 characters => HTTP 400 (bad request)

(No I did not test every character up to 32444 :P)

In other words: There's a ~8 KiB request URL limit and a ~32 KiB total request header limit on the serverside, which is quite normal. I have no clue why you get CloudFlare 520 between 8 and 13.5 KiB.

Adding my "wall of text" as well (Pastebin, it should be 14875 ISO-8859-1 characters (ASCII, all but an "é") (note that it's in the 502 range mentioned above). It's mostly just lists and wiki links and always gives me HTTP 502 in the forum, but as you mentioned it works just fine in the comments, I find that strange....

Oh wait, maybe not. Comments use POST and forum uses GET, so I'm fairly sure that is the problem now. Comments and wiki uses POST so no problems there. Blips, Artist notes and Sets use GET and so they got the same problems.

First time I tried the blips with my wall of text I got a visual 502 in preview box, reloading page and doing it again resulted in no visual, yet still getting the same 502, man that's freaky...

On a similar note: Funnily enough non ASCII works everywhere but in comments... The strange part is that e.g. wiki and comments both use POST but clearly different JavaScript methods (but is an old, apparently not so important bug: forum #139372).

Sounds like this is a bit of a mess, but maybe not a critical bug, since you get about 1500 words before e621 goes mad.

parasprite said:
Are you still having issues with this? I haven't been on as much as usual due to finals, but I'm not seeing this behavior right now. I do remember when the ad server was down a while back you could "fix" the page loading issues temporarily by enabling adblock. Might be worth trying (temporarily of course ;) ) to see if it helps.

No, it was just temporary (a few hours tops), but about the same time as OP reported, just thought I'd mention it. I've got an adblocker but e621 is whitelisted :P

BetrayerOfNihil said:
Screenshot of the problem happening: http://i.imgur.com/JI4yqLz.png
Link to the page shown in the screenshot: https://e621.net/post/show/652077

I can see the image now after taking the time to type this message and then going back to that page and refreshing, but when I first tried to view the picture it wouldn't show up. The example image happens to be a really recent one that just happened to have this problem with it, but MUCH older images do this too sometimes. I'm not sure what causes it, and the only thing I can do is look to the source of the image. If it doesn't have a linked source, or if it does but it's not on Fur Affinity, then I'm pretty much forced to click the back button and forget about trying to view the image.

Most of the time refreshing doesn't help, usually unless I wait several minutes before attempting to refresh. But more often than not, it seems refreshing never works no matter how long I try to patiently wait.

Clearing the whole cache and attempting to view the page again has helped sometimes, but it's not a totally reliable solution and it's annoying to have to sign into everything again.

Heh, "data syphon".

Hmm. That's definitely not related to the ad server issues Chessax and I mentioned. I can't reproduce it on Windows 8.1 IE 11 on my own setup, but that doesn't really rule out something relateed to the browser/OS.

• If you can, try loading the page in a different browser (Firefox/Chrome/Opera/etc.). If you can't reproduce it on those browsers let me know, otherwise continue to the next step.
• Flush the DNS cache: The DNS cache is essentially a big address book for matching "Website.com" to "That server you can find in Arizona" and speeds up browsing. Sometimes a bad entry can slow down browsing or cause loading issues on certain websites. Flushing (or deleting) the cache can sometimes fix random issues like this. It's safe to do and it the cache will be recreated automatically. This page shows you how to do it.

Let me know if you have any questions.

Chessax said:

Good stuff

Thanks for all that detailed information. I'm going to forward this to the devs now so they can have a better look at it.

On a similar note: Funnily enough non ASCII works everywhere but in comments... The strange part is that e.g. wiki and comments both use POST but clearly different JavaScript methods (but is an old, apparently not so important bug: forum #139372).

Oh, would you look at that. Yup, confirmed.

I know they have plans to do some big under-the-hood software updates soon (Ruby 1.8.7 -> 1.9.3 is one of them) and text encoding issues have been a major focus area for testing since there have been some changes with how it's handled. I don't know what this particular bug involves, but I'll go ahead and mark this as a current bug on the dev site to help push for a fix for whenever the actual update gets rolled out.

Videos for the Android tablet (Samsung Galaxy Tab 3) keeps freezing. Only videos that are more than 1 minuet actually work, unter that limit the video freezes. :c

moochie123456 said:
Videos for the Android tablet (Samsung Galaxy Tab 3) keeps freezing. Only videos that are more than 1 minuet actually work, unter that limit the video freezes. :c

Based on your favorites, I'll assume you are talking about webm videos. As far as I understand it, the browser just passes the file onto libvpx, which is what is responsible for actually playing the file. If the file is playing at all, there isn't really anything we can do on our end to fix it. It's either a bug on your browser/Android or an issue specific to the file itself (corrupted or doesn't follow specs).

There might be a few things you can try though:

• Click "download" on the post page to directly play the file. It's a bit of a workaround, but it can help with bugs related to embeding the file in a web page (since it essentially bypasses all of that).
• Update your Android. This is likely what will ultimately fix the issue. I'd update your browser as well if you got it from the Play store.
• Try it in a different browser. I have no idea what browser you use or how other Android browsers handle webm, but it's worth a shot. I personally used Dolphin when I ran Android but there are a lot of options to try. If it works fine in that browser, I'd recommend keeping it around just for viewing stubborn webm files, at least until Android/Samsung/et al. gets around to pushing a bug fix for your tablet.
• If you're technically inclined, there may be some debug settings or something you can play with to get it to work (something related to hardware acceleration, streaming, cache, etc.). I can't help you much here as I'm completely out of the loop with Android, but it may be worth a google search.

If it's actually related to flash files, there definitely isn't anything we can do about it as we have even less control over how Flash Player plays things (also, as far as I remember Android Flash Player was officially discontinued, so this may not even apply to you).

parasprite said:
Heh, "data syphon".

Hmm. That's definitely not related to the ad server issues Chessax and I mentioned. I can't reproduce it on Windows 8.1 IE 11 on my own setup, but that doesn't really rule out something relateed to the browser/OS.

• If you can, try loading the page in a different browser (Firefox/Chrome/Opera/etc.). If you can't reproduce it on those browsers let me know, otherwise continue to the next step.
• Flush the DNS cache: The DNS cache is essentially a big address book for matching "Website.com" to "That server you can find in Arizona" and speeds up browsing. Sometimes a bad entry can slow down browsing or cause loading issues on certain websites. Flushing (or deleting) the cache can sometimes fix random issues like this. It's safe to do and it the cache will be recreated automatically. This page shows you how to do it.

More involved stuff

• If you have recurrent problems with slow browsing and partially loaded websites, changing DNS servers may help in some cases. Usually the one you are using by default is from your ISP, but if your ISP's DNS server is crap changing to another one can fix random issues like this. I'd flush the cache before suggesting this, but if you are interested Google and OpenDNS have alternate ones you can use that are free. (you may be able to do this one your phone as well)
• For your phone: I don't know much about Windows Phones as I've never used them, but you may still have some options. When it's on wifi it will take the DNS settings from your wifi/router (your computer would too by default). As a diagnostic test try switching from wifi to cellular (or somebody else's wifi) and see if it helps. If it does you could try changing the DNS server on your Windows Phone (I'm not sure if you can) or your router (depending on model, it may be hardcoded).

Let me know if you have any questions.

I'll try the browser thing as soon as I get back on my computer. Here is a screenshot from my phone that might be useful, it is what I see when I attempt to download an image that doesn't appear on its page. http://i.imgur.com/r1ViYnc.png

Currently my phone is set to accept ALL cookies, so I'm not sure why it would say "cookie too large". Later I'll see what sort of error my desktop returns when trying to download.

BetrayerOfNihil said:
I'll try the browser thing as soon as I get back on my computer. Here is a screenshot from my phone that might be useful, it is what I see when I attempt to download an image that doesn't appear on its page. http://i.imgur.com/r1ViYnc.png

Currently my phone is set to accept ALL cookies, so I'm not sure why it would say "cookie too large". Later I'll see what sort of error my desktop returns when trying to download.

Yeah that definitely shouldn't be happening, but fortunately it should be an easy fix. You may have to do the same steps for your desktop depending on the error you get on there, but this should apply to your phone at least.

TLDR: You probably have a corrupt cookie and you should be able to fix it by simply deleting it. Here's what you should do next:

• Browse in a private tab: This is just a test to verify that the cookies are causing issues and is optional but recommended. If you go to e621.net in a new private tab it should open the page as if you have no cache/cookies. You can verify this by checking to see if you are logged in (you shouldn't be). Check to see if you are still having issues.
• Option 1 - Clear browser data (cache + cookies): Wipe all cookies on all sites. You'll probably be logged out of any sites you are logged into. It's quick, easy, and temporarily inconvenient but it should work. You should be able to do this by going into your preferences/settings and it should be under "clear browser data", "delete cache and cookies", or something similar. If you can't find it let me know what your OS/browser is and I can help you from there.
• Option 2 - Delete just e621-related cookies: I don't know if you can do this on your phone or if it's worth the bother for you, but alternatively you can just delete all cookies with e621.net, e926.net (if you use it), and probably dragonfru.it as well (the ad server here). There may be multiple cookies under slightly different names such as e621.net, www.e621.net, static.e621.net, etc., just delete any that you find.

Hey something weird has been happening recently. Some posts aren't showing up for me until after I try to upload them and then get a "This post already exists" error.

Bug: Escaped tag query dtext doesn't seem to work

Looks like: \tag query

Should be: Something like: \wiki article

To reproduce:

\{{tag query}}

(that was just escaped escaped dtext)

titanmelon said:
Bug: Escaped tag query dtext doesn't seem to work

Looks like: \tag query

Should be: Something like: \wiki article

To reproduce:

\{{tag query}}

(that was just escaped escaped dtext)

Tony recently added support for in-text code, which I personally find much easier to use than trying to escape stuff.

Example: Text text {{tag query}} text text.

Text text `{{tag query}}` text text.

There's also this really stupid workaround I figured out:

{{tag query}}

{[b][/b]{tag query}}

Doesn't help if you quote the comment though, unfortunately.

Not sure if this is a bug, but none of the recent changelogs mention it.

Bug: Using the wiki search bar no longer parses whitespace as underscores

Expected behavior: ???

Actual behavior: When searching for a tag with underscores, not including said underscores returns a blank page/no result for the selected tag

e.g. bigger version at the source (no result)

vs. bigger_version_at_the_source (tag found)

Steps to duplicate:Search for any tag with underscores using spaces instead

-

parasprite said:

Wow, that's pretty useful!

Not sure if this has been mentioned earlier but:
-Some tags are not working correctly
-Blocked or hidden tags are still appearing

UnknownFurryWolf said:
Not sure if this has been mentioned earlier but:
-Some tags are not working correctly
-Blocked or hidden tags are still appearing

I'll address these separately, but they may be related.

-Blocked or hidden tags are still appearing

It looks like you only have one entry, so I'll start by asking you what happens when you do a search for my_little_pony?

-Some tags are not working correctly

What tags specifically? I don't see anything obvious with the few searches I've tried. It helps to make sure that the posts are tagged with the terms you are trying to search/block, because they won't work unless they are actually tagged.

Bug: Note formatting behaves strangely

Expected behavior: ???

Actual behavior:
According to tests done by parasprite (forum #151860), some dtext in notes is buggy.

Most noticeable example happens when trying to expand a section

Colours still don't work afaik (using [color=x][/color])

Steps to duplicate: Try to add colour dtext on any note

Interestingly enough, css formatting seems to work, but only briefly before being reverted to plaintext

Is there a workaround for this?

titanmelon said:
Bug: Note formatting behaves strangely

Expected behavior: ???

I laughed. Yeah, from what I understand dtext formatting on notes is more of a side effect than an intended feature. It's not really meant to handle much more than plain text and the occasional italics as it's almost entirely used for translations (though there are a few cases like post #26455 where it ended up taking an entirely different direction). As you can tell, it was originally written to accept html formatting and there are still some artifacts of that laying around in notes. It works something like this:

• 1. User adds random text in a note, in this case with some [\b]dtext formatting[/b] and hits save
• 2a. The page parses a client-side "preview" on save based on the text in the box (after all, why resend the page again if you don't have to?)
• 2b. A copy of the raw text is also sent to the server which cleans/optimizes it and sticks it in the database
• 3. If all goes well, the preview (2a) and the real note (2b) should look exactly the same

The reality though is that 2b was easy to implement, but never refined that much (which is why you can still use the same section, quote, etc. tags as you can everywhere else). And for 2b, DText was kind of shoved on top of the legacy code as a compatibiity layer, but other than that it really hasn't changed that much.

Comments/forums/blips/wiki on the other hand re written in a way that more/less merges 2a and 2b to simplify and standardize display. Both the preview button and the "psudo-preview" you get when saving a comment/forum/blip/wiki are actually sent to the server, sanitized, parsed and sent back for your browser to display rather than doing it client-side, however even these still aren't quite in sync with each other yet so you get some odd quirks (see disappearing spaces).

Actual behavior:
According to tests done by parasprite (forum #151860), some dtext in notes is buggy.

Most noticeable example happens when trying to expand a section

Can you elaborate? I'm not exactly sure what you mean by this as they seem to expand/collapse just fine for me.

[color]Colours still don't work afaik (using [color=x][/color])

Steps to duplicate: Try to add colour dtext on any note

The color markup is normally priv+ only, and since members can edit them they were purposely disabled. If you'll notice, anywhere that's publicly editable (the wiki, description fields, etc.) all have the color markup disabled.

Interestingly enough, css formatting seems to work, but only briefly before being reverted to plaintext

Is there a workaround for this?

The CSS formatting quirks you see are just unintentional artifacts of the old code. The only recommendation I have is to stick with standard DText as any other markup will be disabled some time in the near future, as it is on other parts of the website (comments, wiki, etc.).

Bug: The Atom feed is not working again.

Expected behavior: The Atom feed should add entries when art is posted.

Actual behavior: The Atom feed has been empty for months.

Steps to duplicate: Add the atom feed to a feed reader and expect posts, but receive none.

I have reported this bug three times.

It has been fixed three times.

Now it is broken again.

Would it be possible to add some sort of routine check to verify that it isn't broken, or to find the source of the breakage and patch it?

The self-signed Java IRC applet introduces a significant security vulnerability.

Your notice on the help page reads:

> you trust us not to give you malware (we won't!)

This implies that the only possible vector for an exploit would be for you to inject malware into the Java applet.

This is a false pretense.

It does not account for the possibility of cross site scripting, network intercept of private data, injection of code from a third party via unknown exploits or any of dozens of other common and easy to find exploits in Java and the site's code.

The applet should either be correctly signed or removed.

At the very least the help document should be updated to make the user fully aware of the gravity of their decision to run unsigned Java code on their machine.

Dason said:
Bug: The Atom feed is not working again.

Expected behavior: The Atom feed should add entries when art is posted.

Actual behavior: The Atom feed has been empty for months.

Steps to duplicate: Add the atom feed to a feed reader and expect posts, but receive none.

I have reported this bug three times.

It has been fixed three times.

Now it is broken again.

Would it be possible to add some sort of routine check to verify that it isn't broken, or to find the source of the breakage and patch it?

I didn't even know we had an atom feed, hmm... Ah yes there it is. https://e621.net/post/atom/

I tried adding that particular URL to a couple different feed readers and it seemed to pull 32 results each time. https://e621.net/post/atom?tags=cat pulled up a cat feed, and so on. What reader are you using?

Dason said:
The self-signed Java IRC applet introduces a significant security vulnerability.

Your notice on the help page reads:

> you trust us not to give you malware (we won't!)

This implies that the only possible vector for an exploit would be for you to inject malware into the Java applet.

This is a false pretense.

It does not account for the possibility of cross site scripting, network intercept of private data, injection of code from a third party via unknown exploits or any of dozens of other common and easy to find exploits in Java and the site's code.

The applet should either be correctly signed or removed.

At the very least the help document should be updated to make the user fully aware of the gravity of their decision to run unsigned Java code on their machine.

We are aware of the known security problems with trusting self-signed certificates, which is why we strongly recommend you use a dedicated IRC client instead.

Truth is, we've been planning on replacing Java applet for a while now. Even not considering the secuirty problems the Java plugin has, it's becoming less and less common to see users with it installed at all, either by choice or lack of need for it. Getting it correctly signed is expensive when compared next to the small number of users that actually use it. Even if we got it correctly signed, we'd still recommend you use a dedicated client as they have a lot of benefits over running in an applet anyways. We've got a promising candidate in testing right now but I don't have a timeframe to give for when it will all get implemented.

That being said, I will be bringing your concerns to the attention of the devs to see if we can find a better way of handling it in the meantime.

The pool seems to be shuffling itself. I refreshed the page a few times, each time making what was on the screen different.

Ray_Clocks said:
The pool seems to be shuffling itself. I refreshed the page a few times, each time making what was on the screen different.

Do you have a link to a pool that is doing this?

Dason said:
The self-signed Java IRC applet introduces a significant security vulnerability.

The applet should either be correctly signed or removed.

At the very least the help document should be updated to make the user fully aware of the gravity of their decision to run unsigned Java code on their machine.

A replacement (Mibbit) is in the next update, now that the IRC network we use now allows Mibbit clients.

Bug: Blacklist aliased length processing contains non matching regex "/(\d+) \//" when matching string "? / 3900" (as well as other problems)
Expected behavior: Fully aliased blacklist length is displayed?
Actual behavior: Unhandled JavaScript error (null access)
Steps to duplicate: Type anything in blacklist

In addition the code doesn't seem to make any sense, it just seems to spit out question marks, based of how many digits in the match (which is completely pointless):

digits = $('aliased-blacklist-length').innerHTML.match(/(\d+) \//)[1].length
q = ""
for (i = 0; i < digits; i++) {
 q += "?"
}
$('aliased-blacklist-length').innerHTML = q + " / 3900"

And the "Update" AJAX link (checkBlacklistLength()) produces HTTP 500 due to authentication token problems. I don't know, but it looks like this whole thing was an attempt to preview/test a feature that never got fully implemented:

Invalid auth token!If you came here from another site,you were probably just subject to a cross-site request forgery attack!

However, messing around, I tried the URL directly with https://e621.net/user/blacklist_length?blacklist=barbed_cock%0d%0aass, and resulted in what appears to be the proper aliased length as far as I know it, i.e. "barbs+butt" => 5+1+4=10 characters, so some parts seem to work.

Don't work on android

Chessax said:
Bug: Blacklist aliased length processing contains non matching regex "/(\d+) \//" when matching string "? / 3900" (as well as other problems)
Expected behavior: Fully aliased blacklist length is displayed?
Actual behavior: Unhandled JavaScript error (null access)
Steps to duplicate: Type anything in blacklist

In addition the code doesn't seem to make any sense, it just seems to spit out question marks, based of how many digits in the match (which is completely pointless):

And the "Update" AJAX link (checkBlacklistLength()) produces HTTP 500 due to authentication token problems. I don't know, but it looks like this whole thing was an attempt to preview/test a feature that never got fully implemented:

However, messing around, I tried the URL directly with https://e621.net/user/blacklist_length?blacklist=barbed_cock%0d%0aass, and resulted in what appears to be the proper aliased length as far as I know it, i.e. "barbs+butt" => 5+1+4=10 characters, so some parts seem to work.

Just fixed it for the next update, thanks! The whole thing didn't work because when we implemented the authenticity token system (to prevent CSRF attacks) which protects all /user/ pages, we forgot about that call.

The question mark code simply replaces 7 / 3900 with ? / 3900 or 257 / 3900 with ??? / 3900, indicating that the aliased length is unknown until you press Update.

hhstrane said:
Don't work on android

There may be a number of different reasons for this as Android has hundreds of versions and setups to go off of, so I'll need a bit of help narrowing it down:

• What specifically doesn't work?
• Do you happen to be using the browser Dolphin?
• What is the name of your device? Version of Android? (if you know how to find it)

The count for the search results is inconsistent (For me, the search results do not 'scale', evenly, from page to page.

TonyLemur said:
Just fixed it for the next update, thanks! ... The question mark code simply replaces 7 / 3900 with ? / 3900 ... until you press Update.

My pleasure. Though, thinking back I probably overreacted a bit, it makes more sense now, just had a "what the fudge were they thinking"-moment.

chaccaronmacarondo said:
The count for the search results is inconsistent (For me, the search results do not 'scale', evenly, from page to page.

Just a thought: Do you happen to have anything blacklisted? The blacklist is currently applied client-side only so if the server gives you 75 thumbnails and 25 of those match your blacklist, only 50 will be visible.

parasprite said:
There may be a number of different reasons for this as Android has hundreds of versions and setups to go off of, so I'll need a bit of help narrowing it down:

• What specifically doesn't work?
• Do you happen to be using the browser Dolphin?
• What is the name of your device? Version of Android? (if you know how to find it)

Sorry I'm not disable jetpack

hhstrane said:
Sorry I'm not disable jetpack

I'm sorry, but do you mean that you refuse to disable jetpack? You're free to keep it enabled, but from what I understand it is actually jetpack itself that causes the issues (a lot of sites have reported similar problems with jetpack); I don't know that there is much we can do on our end to work around it.

We will keep looking for workarounds, but in the meantime we still recommend users try disabling jetpack first.

chaccaronmacarondo said:
The count for the search results is inconsistent (For me, the search results do not 'scale', evenly, from page to page.

Chessax said:
Just a thought: Do you happen to have anything blacklisted? The blacklist is currently applied client-side only so if the server gives you 75 thumbnails and 25 of those match your blacklist, only 50 will be visible.

Looking at your blacklist, I'd say this would definitely cause behavior like this. If you haven't already, you can minimize this effect by going to your settings page and increasing Posts per page to 150 (or more if you prefer).

Peekaboo said:
Uploads are broken again.
Edit: Not the uploads that were broken, it's just that using the pool: metatag to add a post to a comic in the upload causes the Action controller error.

Not sure if this is the same problem but...
Bug: "Add to pool" dropbox isn't showing every pool

Expected behavior: "Add to pool" will show every pool currently existing. I was able to see them all only a few days ago.

Actual behavior: "Add to pool" is only showing 12 pools, and is slowly showing a few more every day. It started showing none at first, but as the days pass it shows more. I don't think waiting this long for my preferred pool to show up is ideal. It started doing this when I uploaded this post I even made another account to see if my current one was the issue, but the same problem presents itself with the new account.

Steps to duplicate: Attempt to add a post to a pool.

Ghost524 said:
Not sure if this is the same problem but...
Bug: "Add to pool" dropbox isn't showing every pool

Expected behavior: "Add to pool" will show every pool currently existing. I was able to see them all only a few days ago.

Actual behavior: "Add to pool" is only showing 12 pools, and is slowly showing a few more every day. It started showing none at first, but as the days pass it shows more. I don't think waiting this long for my preferred pool to show up is ideal. It started doing this when I uploaded this post I even made another account to see if my current one was the issue, but the same problem presents itself with the new account.

Steps to duplicate: Attempt to add a post to a pool.

Nope, it's not related. It's actually not a bug either.

That list contains all pools currently marked "active". Up until about a week ago there were thousands of entries in that list that had accumulated over time that you had to scroll endlessly through to find the pool you were looking for. Since having thousands of mostly finished pools in that list is kind of pointless and confusing for many people (some even thought it was a way to browse pools), I went through and cleaned the list out manually; the pools you see are probably going to all be new pools (new pools are marked active by default). This was kind of a one time thing and I doubt it will be done very often in the future, but if it does I think leaving a chunk of new pools (the last 100? 500?) should suffice.

If you want to add posts to a pool that isn't in that list you can do any of the following:

• Tag it with pool:XXXX. In other words, for pool #1234 you'd tag it with pool:1234 (doesn't need to be active) OR
• Go to the pool itself and change it to active, then use the add to pool list OR
• Use the import function — great for comics (doesn't need to be active)
• Tip: I'd also recommend adding order:id_asc to your import search since this will show the correct order more often than not

Edit: If you happen to know the exact name, you can also tag it with that:

• pool:mlp:_my_little_dragon for pool #1234 - Note that this may be a bit hit/miss since it has to be exactly the same as on the edit page.

If this problem is known just ignore me:

Bug: Uploading swf files results in an error trace

Expected behavior: Uploading and saving the flash file to the site

Actual behavior: Getting redirected to a page where it shows this error:

NoMethodError in PostController#create

undefined method `>' for nil:NilClass

along with this Application Trace:

/home/e621/e621-production/release-2014-06-08/app/models/post/file_methods.rb:182:in `set_image_dimensions'
/home/e621/e621-production/release-2014-06-08/app/controllers/post_controller.rb:59:in `create'

Steps to duplicate: Upload a swf file

Ladon_Dragon said:
If this problem is known just ignore me:

Bug: Uploading swf files results in an error trace

Expected behavior: Uploading and saving the flash file to the site

Actual behavior: Getting redirected to a page where it shows this error:

NoMethodError in PostController#create

undefined method `>' for nil:NilClass

along with this Application Trace:

/home/e621/e621-production/release-2014-06-08/app/models/post/file_methods.rb:182:in `set_image_dimensions'
/home/e621/e621-production/release-2014-06-08/app/controllers/post_controller.rb:59:in `create'

Steps to duplicate: Upload a swf file

I don't have the ability to test swfs right now, but it looks like it may be having trouble with the file itself, so it might have been corrupted on download or at the source. It's worth mentioning that you can also upload things by leaving the "chose file" blank and putting the direct link in the source field instead (note that this won't work for certain sites like inkbunny which require you to log in).

If it's still giving you the same error using that, please send me a copy of the file to look at.

parasprite said:
Nope, it's not related. It's actually not a bug either.

That list contains all pools currently marked "active". Up until about a week ago there were thousands of entries in that list that had accumulated over time that you had to scroll endlessly through to find the pool you were looking for. Since having thousands of mostly finished pools in that list is kind of pointless and confusing for many people (some even thought it was a way to browse pools), I went through and cleaned the list out manually; the pools you see are probably going to all be new pools (new pools are marked active by default). This was kind of a one time thing and I doubt it will be done very often in the future, but if it does I think leaving a chunk of new pools (the last 100? 500?) should suffice.

If you want to add posts to a pool that isn't in that list you can do any of the following:

• Tag it with pool:XXXX. In other words, for pool #1234 you'd tag it with pool:1234 (doesn't need to be active) OR
• Go to the pool itself and change it to active, then use the add to pool list OR
• Use the import function — great for comics (doesn't need to be active)
• Tip: I'd also recommend adding order:id_asc to your import search since this will show the correct order more often than not

Well shit. I learned a lot today. 'Preciate it!

parasprite said:
I don't have the ability to test swfs right now, but it looks like it may be having trouble with the file itself, so it might have been corrupted on download or at the source. It's worth mentioning that you can also upload things by leaving the "chose file" blank and putting the direct link in the source field instead (note that this won't work for certain sites like inkbunny which require you to log in).

If it's still giving you the same error using that, please send me a copy of the file to look at.

Trying to leave it empty and loading it from the source (from my own webserver) just gives me:

Error: File is an invalid content type: 

The file for download can be found here in case you want to look into it (it's the only file there)(also it's NSFW):
http://ladon-dragon.net/downloads/megapoly/

(I did use the full link to the swf in the source)

Ghost524 said:
Well shit. I learned a lot today. 'Preciate it!

Is an admin needed to make the pool active? I can't find the option.

Ladon_Dragon said:
Trying to leave it empty and loading it from the source (from my own webserver) just gives me:

Error: File is an invalid content type: 

The file for download can be found here in case you want to look into it (it's the only file there)(also it's NSFW):
http://ladon-dragon.net/downloads/megapoly/

(I did use the full link to the swf in the source)

I tried using a third party SWF viewer and apparently this file was not supported (and it plays just about any other flash (never ever seen that error before)), so my guess is that you got a too new or too strange file format or a somehow slightly corrupted file (however playing it in Chrome as well as in Adobe Flash Player worked). I'm curious to know what software you used to create the file.

Smudge_Proof said:
Is an admin needed to make the pool active? I can't find the option.

Simply go to the pool, or click Pools at the the top menu and search for it, then just under the top menu you got an "Edit" link, if that isn't there the pool is private and you have to ask the creator or an admin (or similar) to do it for you.

Chessax said:
I tried using a third party SWF viewer and apparently this file was not supported (and it plays just about any other flash (never ever seen that error before)), so my guess is that you got a too new or too strange file format or a somehow slightly corrupted file (however playing it in Chrome as well as in Adobe Flash Player worked). I'm curious to know what software you used to create the file.

Simply go to the pool, or click Pools at the the top menu and search for it, then just under the top menu you got an "Edit" link, if that isn't there the pool is private and you have to ask the creator or an admin (or similar) to do it for you.

It was created using Flash CC

Ladon_Dragon said:
If this problem is known just ignore me:

Bug: Uploading swf files results in an error trace

Expected behavior: Uploading and saving the flash file to the site

Actual behavior: Getting redirected to a page where it shows this error:

NoMethodError in PostController#create

undefined method `>' for nil:NilClass

along with this Application Trace:

/home/e621/e621-production/release-2014-06-08/app/models/post/file_methods.rb:182:in `set_image_dimensions'
/home/e621/e621-production/release-2014-06-08/app/controllers/post_controller.rb:59:in `create'

Steps to duplicate: Upload a swf file

I can't find it in our logs but I believe we determined that that error was due to a non-standard compression method that our libraries don't support.

TonyLemur said:
I can't find it in our logs but I believe we determined that that error was due to a non-standard compression method that our libraries don't support.

So a fix would be trying to change the compression method in Flash CC and try again?

Smudge_Proof said:
Is an admin needed to make the pool active? I can't find the option.

https://e621.net/pool/update/6104 - Second box

Ladon_Dragon said:
So a fix would be trying to change the compression method in Flash CC and try again?

I'd say it's worth trying. I haven't used Flash CC before, but if there are some compatibility option for older versions I'd use that.

Chessax said:
I tried using a third party SWF viewer and apparently this file was not supported (and it plays just about any other flash (never ever seen that error before)), so my guess is that you got a too new or too strange file format or a somehow slightly corrupted file (however playing it in Chrome as well as in Adobe Flash Player worked). I'm curious to know what software you used to create the file.

VLC just thinks there's an audio channel. :V

Ladon_Dragon said:
So a fix would be trying to change the compression method in Flash CC and try again?

Did some testing myself with both ZLIB (Deflate) and LZMA compression. Deflate seems to behave as expected everywhere, LZMA not so much. The best solution is probably to go with "Deflate" at the moment and try again.

TonyLemur said:
I can't find it in our logs but I believe we determined that that error was due to a non-standard compression method that our libraries don't support.

Might be something to put on a todo list in case LZMA is not supported, as it is now a standard, but not default, compression option in Adobe Flash software. It's probably going to get more and more common.

parasprite said:
VLC just thinks there's an audio channel. :V

Have to try that... *disappointment* I expected at least some kind of sound :(

parasprite said:
https://e621.net/pool/update/6104 - Second box

I'd say it's worth trying. I haven't used Flash CC before, but if there are some compatibility option for older versions I'd use that.

VLC just thinks there's an audio channel. :V

Chessax said:
Did some testing myself with both ZLIB (Deflate) and LZMA compression. Deflate seems to behave as expected everywhere, LZMA not so much. The best solution is probably to go with "Deflate" at the moment and try again.

Might be something to put on a todo list in case LZMA is not supported, as it is now a standard, but not default, compression option in Adobe Flash software. It's probably going to get more and more common.

Have to try that... *disappointment* I expected at least some kind of sound :(

Reporting back:

Uploading a SWF that is compressed using Deflate works, thanks!

https://e621.net/post/show/661055/anal-anal_penetration-anthro-anus-balls-blue_eyes-

I want to add this to its respective pool, but the Add To Pool menu seems to be broken. Only about 25 of them show up on the list.

Jugofthat said:
https://e621.net/post/show/661055/anal-anal_penetration-anthro-anus-balls-blue_eyes-

I want to add this to its respective pool, but the Add To Pool menu seems to be broken. Only about 25 of them show up on the list.

The "add to pool" list was recently cleaned out as it had thousands of entries in it that accumulated over the years. To add to pools not on that list you can either tag it with pool:1234 (using pool ID) or go to the edit page in that pool and mark it as "active" (second box). Marking it as active will add it to that list.

I see. Could be a slight hindrance when people want to add new posts to an older pool and every one of them needs to be reactivated first, but I understand the reasoning behind it, and reactivating a pool is easy when you know how. :)

Thing is, obviously not everyone will know. Perhaps you guys may wanna put this info in the News bracket at the top for a few days, to prevent more of this confusion (and repeat questions).

That sound decent?

Bug
Loggin in to e621 doesn't automatically log me in to e926

Expected behavior
When I'm logged in to https://e621.net/ I should also be logged in to http://www.e926.net/
(not necessarily vice versa, though)

Actual behavior
I'm logged in to e621, but not to e926

Steps to duplicate
Log in to e621 and go to e926

Munkelzahn said:
Bug
Loggin in to e621 doesn't automatically log me in to e926

Expected behavior
When I'm logged in to https://e621.net/ I should also be logged in to http://www.e926.net/
(not necessarily vice versa, though)

Not possible due to how cookies work (they cannot be shared across domains).

Well, technically possible by redirecting the user back and forth between sites when they log in, but that's kind of messy and annoying and in my opinion not worth the minor convenience.

Blacklist Update button doesn't work anymore. i know i am typing everything in right, but it refuses to update the blacklist.

Trying to add (also tried a handful of random tags to verify)

rating:explicit

so i can show my sister some of the art, without getting bombarded with dongs,

FeedTheFurry said:
Blacklist Update button doesn't work anymore. i know i am typing everything in right, but it refuses to update the blacklist.

Trying to add (also tried a handful of random tags to verify)

rating:explicit

so i can show my sister some of the art, without getting bombarded with dongs,

You can go to http://e926.net for that (you'll have to login there too). That won't fix the bug though, but it's the SFW alternative of e621

I am getting an odd bug at seemingly random times. What's happening is on the first page it will show like 10 pictures, second page shows 1, third shows me 4, fifth shows me 7, sixth shows 2 etc etc. It doesn't seem to matter what tags I use either.

Bug: Tag history API not recognizing date_after/date_before field as described. It might be a simple case of not knowing proper date format, tried various but assumed Unix timestamp. Pretty minor thing, but thought I'd mention it.
Expected behavior: Fields are honored
Actual behavior: Fields are ignored
Steps to duplicate: E.g. this should return a single tag change, my first at 1380287944 :
https://e621.net/post_tag_history/index.xml?user_id=89724&date_before=1380287940

Xenoifsa said:
I am getting an odd bug at seemingly random times. What's happening is on the first page it will show like 10 pictures, second page shows 1, third shows me 4, fifth shows me 7, sixth shows 2 etc etc. It doesn't seem to matter what tags I use either.

As a control, try searching for -my_little_pony and tell me what you see.

FeedTheFurry said:
Blacklist Update button doesn't work anymore. i know i am typing everything in right, but it refuses to update the blacklist.

Trying to add (also tried a handful of random tags to verify)

rating:explicit

so i can show my sister some of the art, without getting bombarded with dongs,

The "update" button just updates the character count. You have to hit save at the bottom of the page in order for it to save the blacklist.

Jugofthat said:
I see. Could be a slight hindrance when people want to add new posts to an older pool and every one of them needs to be reactivated first, but I understand the reasoning behind it, and reactivating a pool is easy when you know how. :)

Thing is, obviously not everyone will know. Perhaps you guys may wanna put this info in the News bracket at the top for a few days, to prevent more of this confusion (and repeat questions).

That sound decent?

I wish I would have thought of doing this a week ago. :/

Chessax said:
Bug: Tag history API not recognizing date_after/date_before field as described. It might be a simple case of not knowing proper date format, tried various but assumed Unix timestamp. Pretty minor thing, but thought I'd mention it.
Expected behavior: Fields are honored
Actual behavior: Fields are ignored
Steps to duplicate: E.g. this should return a single tag change, my first at 1380287944 :
https://e621.net/post_tag_history/index.xml?user_id=89724&date_before=1380287940

I'm on a mobile device so my ability to test things is limited, but a few things you could try:

• "year-month-day" format (e.g., 2015-05-27) - I believe JSON only allows this format (despite giving you Unix time) but I'm not sure about XML since I rarely use it
• Try using both date_before and date_after - It may require both, but I haven't played with it enough to tell
• Try using date_start and date_end instead - I remember something on here used this, but I could be wrong (they may also both work :V)

parasprite said:
I'm on a mobile device so my ability to test things is limited, but a few things you could try:

• "year-month-day" format (e.g., 2015-05-27) - I believe JSON only allows this format (despite giving you Unix time) but I'm not sure about XML since I rarely use it
• Try using both date_before and date_after - It may require both, but I haven't played with it enough to tell
• Try using date_start and date_end instead - I remember something on here used this, but I could be wrong (they may also both work :V)

Great, date_start and date_end seems to work with simple dates (e.g. 2014-10-01/20141001), date_start inclusive, date_end exclusive and only one is needed. But then the API instructions are wrong, also it would be nice if it would actually take an ISO 8601 date e.g. 2014-10-01T23:59:59+00:00 or simply a Unix timestamp, since dates are clearly stored with high enough precision.

FeedTheFurry said:
Blacklist Update button doesn't work anymore. i know i am typing everything in right, but it refuses to update the blacklist.

Trying to add (also tried a handful of random tags to verify)

rating:explicit

so i can show my sister some of the art, without getting bombarded with dongs,

Answering again for visibility (sorry Parasprite :P):

your blacklist still works fine, assuming you click 'Save' at the bottom of the page to actually save it. The update button (which will be fixed in the next update) only updates the final character count of the blacklist after applying aliases to it.

For example, typing in mlp will actually use 14 characters since it'll be converted to my_little_pony after you save it.

Chessax said:
Bug: Tag history API not recognizing date_after/date_before field as described

I renamed the params to date_start/date_end but forgot to update the API help page. Just fixed that and made it use date_before/date_after, if they exist and date_start/date_end don't.

Chessax said:
also it would be nice if it would actually take an ISO 8601 date e.g. 2014-10-01T23:59:59+00:00 or simply a Unix timestamp, since dates are clearly stored with high enough precision.

It should currently accept ISO 8601 dates (along with most other common date formats, such as "April 25, 2015"), but now it'll accept Unix timestamps as well. As an added bonus, it'll also flip the dates around if you get them backwards (i.e. April 27 to April 25).

As always, these changes aren't live yet; they'll be made live in the next update.

TonyLemur said:
It should currently accept ISO 8601 dates (along with most other common date formats, such as "April 25, 2015"), but now it'll accept Unix timestamps as well. As an added bonus, it'll also flip the dates around if you get them backwards (i.e. April 27 to April 25).

Really? I guess I must be doing something wrong then (as usual), the following query gives tag changes during 24h and not 1s, it gives 408 results when there should be 0, but maybe you meant it accepts dates and not combined dates with time?:
https://e621.net/post_tag_history/index.xml?limit=1000&date_start=2010-01-01T23:59:59Z&date_end=2010-01-02T00:00:00Z

Chessax said:
Really? I guess I must be doing something wrong then (as usual), the following query gives tag changes during 24h and not 1s, it gives 408 results when there should be 0, but maybe you meant it accepts dates and not combined dates with time?:
https://e621.net/post_tag_history/index.xml?limit=1000&date_start=2010-01-01T23:59:59Z&date_end=2010-01-02T00:00:00Z

Oh, I just remembered that I also modified the method used to parse input dates to handle sub-day precision, so yeah, that won't work until the update. Sorry!

Daigaijin's Better Late Than Never pool is messed up (possibly other pools). When I click on >> to go to the next page in the pool, it goes to a seemingly random page further back in the pool. I'm not sure about << but the >> doesn't work.

Edit: The << on a page acts like the >> does as well.

AshuraK said:
Daigaijin's Better Late Than Never pool is messed up (possibly other pools). When I click on >> to go to the next page in the pool, it goes to a seemingly random page further back in the pool. I'm not sure about << but the >> doesn't work.

Edit: The << on a page acts like the >> does as well.

Ouch, someone managed to wipe the ordering on it. Every post was set to #0. I'm not exactly sure how the server does ordering when there's a conflict like that, but setting every post to #0 would definitely screw up the ordering.

Thankfully the revert button managed to fix it without needing to putz around with it. :P

parasprite said:
Ouch, someone managed to wipe the ordering on it. Every post was set to #0. I'm not exactly sure how the server does ordering when there's a conflict like that, but setting every post to #0 would definitely screw up the ordering.

Thankfully the revert button managed to fix it without needing to putz around with it. :P

Do you know Reiger's ♥ comic (pool #903)? Once I went through the whole thing to manually fix ordering issues. Even that took a ridiculous amount of time. Now if something like BLTN needed a complete manual re-ordering... I wouldn't even touch that with a stick.

EsalRider said:
Do you know Reiger's ♥ comic (pool #903)? Once I went through the whole thing to manually fix ordering issues. Even that took a ridiculous amount of time. Now if something like BLTN needed a complete manual re-ordering... I wouldn't even touch that with a stick.

Reiger's ♥ comic was a bit of a pet tagging project; it needed a lot of work. :x

I actually went through BLTN recently to match the page # displayed on the image/on the artist's inkbunny semi-recently. Looks like it's still good, which is fortunate because you're absolutely right, that is not a pleasant task.

Tags on this submission https://e621.net/post/show/603636/ are broken.
"clothed", "clothing" and "half_dressed" refuse to be deleted no matter what. The tag history will show them being deleted, but the submission will still feature them. Then any unrelated new edit to the tags will show them being re-added in the tag history.

It's really weird, never seen that happening with any other submission.

metatron said:
Tags on this submission https://e621.net/post/show/603636/ are broken.
"clothed", "clothing" and "half_dressed" refuse to be deleted no matter what. The tag history will show them being deleted, but the submission will still feature them. Then any unrelated new edit to the tags will show them being re-added in the tag history.

It's really weird, never seen that happening with any other submission.

That is because it also had the "bottomless" tag, which implies half-dressed. I removed the bottomless tag, so the problem is fixed.

Clawdragons said:
That is because it also had the "bottomless" tag, which implies half-dressed. I removed the bottomless tag, so the problem is fixed.

Ah, interesting, wasn't aware tags could pull other tags. Thanks!

Bug: Did someone break tag and description history?
Expected behavior: Works
Actual behavior: NoMethodError: undefined method match' for nil:NilClass [b]Steps to duplicate:[/b] E.g.: https://e621.net/post_tag_history/index?post_id=13 https://e621.net/post_desc_history/index?post_id=13

Chessax said:
Bug: Did someone break tag and description history?
Expected behavior: Works
Actual behavior: NoMethodError: undefined method match' for nil:NilClass [b]Steps to duplicate:[/b] E.g.: https://e621.net/post_tag_history/index?post_id=13 https://e621.net/post_desc_history/index?post_id=13 [/quote] Oops. Spight pushed everything live last night, but there's an error causing it to not work when no dates are supplied (such as just viewing the index). It will work if you supply both start and end dates, however. Fix should be made live later today. Sorry!

TonyLemur said:
Oops. Spight pushed everything live last night, but there's an error causing it to not work when no dates are supplied (such as just viewing the index). It will work if you supply both start and end dates, however. Fix should be made live later today. Sorry!

It will be fixed soon anyways, but if you happen to need it in the meantime just add &date_start=&date_end= to the end of the url like so:

https://e621.net/post_tag_history?post_id=13&date_start=&date_end=
https://e621.net/post_desc_history?post_id=13&date_start=&date_end=

If it doesn't work (such as when viewing the index page) use ?date_start=&date_end= instead.