It seems that anything that returns "++" (EDIT: in the "About" box), in this case "Notepad++", breaks their internal regex because it seems not to be sanitized properly

The breaking code using chrome dev tools seems to be:

n = new RegExp("^(https?://)?([\\w\\.-_]\\.)?" + o.toLowerCase().replace(/[-\s]/g, "") + "\\.");

In this case "o" is "Notepad++" which might be the first result.

The browser returns this error:

Invalid regular expression: /^(https?://)?([\w\.-_]\.)?notepad++\./: Nothing to repeat

Also happens with C++ https://duckduckgo.com/?q=c%2B%2B&t=ffab&ia=web

Speaking of control characters breaking things, HN ate the * characters in your post and turned the text into italics instead.

its because of markdown formatting

Corroborating this to some extent, the “HTML” and “lite” versions works fine in both cases:

https://html.duckduckgo.com/html/?q=notepad

https://html.duckduckgo.com/html/?q=c%2B%2B

https://lite.duckduckgo.com/lite/?q=notepad

https://lite.duckduckgo.com/lite/?q=c%2B%2B

The part of regex

    [\w\.-_]

should really be

    [\w\.\-_]

instead.

But not that's not what caused the problem mentioned in this post.

(edited for clairifcation)

Curious that the error does not occur when searching for "Notepad+": https://duckduckgo.com/?q=notepad%2B&t=ffab&atb=v101-1&ia=we...

In this scenario all the "o" instances are "Notepad+".

The error seems to be coming from the right sidebar which displays "Notepad+" as a title as well. It seems to be getting the "Heading" from the wikipedia page

Here is the object where it extracts the heading from, it seems in this scenario it is not extracting "Notepad++" in the about but "Notepad+" which is another application.

```

Abstract: "Notepad+ is a freeware text editor for Windows operating systems and is intended as a replacement for the Notepad editor installed by default on Windows. It has more formatting features but, like Notepad, works only with plain text. It can open text files of any size, and a single instance of the program can have multiple files open simultaneously. It supports dragging and dropping text within a file and between files, and supports multiple fonts and colours. Notepad+ is available from the company RogSoft. It was developed by Dutch programmer Rogier Meurs. It was first released in 1996. Originally, it had the advantage of being able to open files of any size, because until 2000 Notepad could not open files larger than 64 KB." AbstractSource: "Wikipedia" AbstractURL: "https://en.wikipedia.org/wiki/Notepad%2B" Heading: "Notepad+"

```

It's because a single plus '+' is valid in a regex. A double plus '++' is invalid.

It's not quite anything that returns a '++', but I think any search term that creates a '++' (or perhaps any invalid regex syntax) inside the main info box result.

"cpp" seems to work fine though, and returns many results that include "c++"... https://duckduckgo.com/?q=cpp&t=ffab&ia=web

But "c plus plus" breaks

Hi all (CEO & Founder, DuckDuckGo here). Looks like something may have broken in a release that went out today -- it is triggering a JS error. We're looking into it now and should have it resolved soon.

UPDATE: this should be fixed now.

Fixed for me (UK), would be interesting to hear what the error was. I did so love the time when you were developing DDG and posting to HN about your progress and getting suggestions from the community and whatnot. So glad it worked out for you.

Thanks! It was a regex error.

Similarly to DNS, its always regex.

Why doesn't DDG return consistent results?

If I search for something, I'll get a list of results, potentially even a featured wikipedia entry or whatnot on the right. If I click on one of the top ranked links and realize it's not what I want, the results page is completely different when I return to it via the back button. e.g. just now I wanted to go back and click on that featured wikipedia link but it's gone now.

FWIW this is the sort of behavior that means I just append '!g' to most queries.

Thanks for acknowledging it and sharing helpful info. It seemed bizarre for a short moment.

Hello! Why has DDG gotten so worse lately? I thought it was just me but I keep reading the same experience on HN. There is no way you don't know about this?

Asking as a long time user and at a time even DDG evangelist.

Cheers!

Probably because they've been piggybacking off other's search indexes for a while, and the Western World seems downright opposed to actually providing an accurate index of the actual content of the Internet.

There's only so much you can do when your primary index builder's search indexes are probably more accurately reconstituted from https://www.lumendatabase.org/lumendatabase

than their live index.

But hey, that's the price of innovation, right? Can't denormalize access to frigging anything. Have to maintain the info asymmetry.

I signed up for Kagi finally. Still figuring out if it's worth it, but the more important thing is that I was moved to even consider it and do it.

The mighty yegg does not answer such questions.

But if you must know: it's talent exodus coupled with deteriorated market share.

I feel like the crowd of e-rubberneckers that gathered here made it seem like a bigger bug than it was...

Using POST, HTTP/1.0 and lite.duckduckgo.com, with no Javascript of course, I got results. Not suggesting any of those are requirements. I'm just reporting that the no results experience is not universal.

Here is the first page of results:

  https://notepad-plus-plus.org/
  https://onlinenotepad.org/notepad
  https://apps.microsoft.com/store/detail/9MSMLRH6LZF3?hl=en-us&gl=US
  https://notepad-plus-plus.org/downloads/v8.5/
  https://sourceforge.net/projects/notepadplusplus.mirror/
  https://notepad.js.org/
  https://notepad-plus-plus.org/downloads/v8.1.9.3/
  https://notepad-plus-plus.org/downloads/v8.1/
  https://support.microsoft.com/en-us/windows/help-in-notepad-4d68c388-2ff2-0e7f-b706-35fb2ab88a8c
  https://www.techspot.com/downloads/2010-notepad.html
  https://github.com/notepad-plus-plus
  https://www.onlinenotepad.io/
  https://anotepad.com/
  https://www.howtogeek.com/902619/open-notepad-in-windows/
  https://notepad.plus/
  https://community.notepad-plus-plus.org/
  https://www.digitalcitizen.life/start-notepad-windows/
  https://notepad.uptodown.com/windows
  https://www.dictionary.com/browse/notepad
  http://www.my-notepad.net/
  https://notepad.id.uptodown.com/windows
  https://notepad-classic-windows-10.en.softonic.com/

Below is a quick script for DuckDuckGo searching, if only want first page of results. yy025 makes HTTP from URLs. yy046 URL-encodes input, e.g., "+" to %2B. The address 127.22.10.55 is a TLS forward proxy.

         usage: echo query string|1.sh > 1.htm
                firefox ./1.htm
                links -no-connect ./1.htm
    
         #!/bin/sh
         
         h=duckduckgogg42xjoc72x3sjasowoarfbgcmvfimaftt6twagswzczad.onion
         h=lite.duckduckgo.com
         read x;
         x=q=$(echo "$x"|yy046);
         export httpMethod=POST;
         export Content_Type=application/x-www-form-urlencoded;
         export Content_Length=${#x};
         export httpVersion=1.0;
         export Connection=close;
         echo https://$h/lite/|yy025|if sed w/dev/stderr;then 
          echo $x;echo $x >&2;fi \
         |socat stdio,ignoreeof tcp:127.22.10.55:80

For that query, the Chrome developer tools console spits an error like so:

>d.7527c434450544045365.js:1 Uncaught (in promise) SyntaxError: Invalid regular expression: /^(https?://)?([\w\.-_]*\.)?notepad++\./: Nothing to repeat (at

If you set a breakpoint there, the value it is trying to replace is "Notepad++" rather than "notepad", gotten from the Heading of "http://www.notepad-plus-plus.org" further up in the logic chain. Looks like the plus signs are incorrectly being added to the RegEx

Was someone trying to be funny, to change a "notepad" search for "notepad++", forgetting that + has a specific meaning in regex? Same error occurs with a search for "notepad++", though, but not "notepad+", funnily enough.

wouldn't this be worse for not having the user provided search query sanitized for special characters?

My assumption is wrong. As others have noted, it's clearly parsing the title of the first result wrong. Edit: It's the parsing of the title of the infobox.

Looks like a regex to detect notepad++ in a url that didn't escape the +'s? Weird, maybe a sponsorship thing

Yep they need to escape every character in the string not just the plus signs

I feel like Google has the same issue. Nowadays, searching for anything on Google yields between 3 and 17 results. Few years ago, I received millions of results for the same searches.

Notepad++ search yield: About 26.000.000 results (0,33 seconds)

so does "cplusplus"

It seems to have to do with some kind of metadata is fetched and then used as a regular expression (to change markup in some other part of the page?); it's assumed that the text can be used as a regular expression without quotation but in the case of the doubled "+" character it becomes a non-valid RE.

Seems to happen with any query containing "otepad".

As does "tepad" though "epad" does return results.

Unless you type in something like Windows Notepad or paper notepad, which does return results.

Strange, very strange.

"Windows os" instead yields just nonsense

What causes this?

They used to have this thing where developers could build Instant Answers through a program called DuckDuckHack.

It would load a carousel at the top with results for like jobs, recipes, etc.

It used regex to match the search query. Looks like they have a bad one.

I have no idea, but equally curious.

Unrelated to this issue, but I get peeved when HN commenters rave how DuckDuckGo results are far superior to Google's, and yet the other day I entered "DuckDNS" on DDG, and it was nowhere to be scene in the top 3 pages. Meanwhile on Google and Brave Search, it is the first result.

If DDG was actually good people would be using it. "It's not Google" doesn't cut it.