newsbin.com

I'm working with v6.50 Build 8, and I'm having problems with the Internet Search.

The initial boolean search seems to work okay (other than lacking the full grouping support I posted about in an earlier thread). However, the followup RegEx search seems to be broken for me. I'm getting results that in no way match what I'm typing.

For example, if you do an initial search for "vivaldi" and then do followup searches such as "g minor" or "op\. ?9" (quotes not part of any of the search patterns), you get completely nonsensical and meaningless results.


Do I recall seeing somewhere that this was deliberate? If so, I'd like to request that RegEx searching be restored to what it should be. If it's not deliberate, then what's going on?

You're using an old beta. I suggest you get current first then try again.

Quade wrote:You're using an old beta. I suggest you get current first then try again.

I thought Internet Search issues were handled by your server rather than the app, but okay. I'll update and try again. Thanks!

You're doing things in both places.

1 - Search and retrieve data. Which happens on the server.

2 - Using the find box to filter down further using regexes. That happens in Newsbin.

From what you're describing, #2 is what you seem to be talking about and that only happens locally.

I just installed the latest beta. It's the same garbage results. It's not properly processing the local search as RegEx.

"g\sminor"

meaning "g" followed by a space followed by the word "minor"

"g minor" just means "g somewhere in the subject AND "minor" someplace in the subject. If you want spaces to be significant, you need to use \s on place of space.
"g.*minor" is almost the same.

So, it's working the way I want it to which is "space = and". You'll need to adjust.

Either it's RegEx or it's not. Hybrid approaches don't work well with the sole exception of adding boolean support.

If you want it to work this way, that's fine - but can you at least add a switch that enables pure RegEx mode?

Edit: I can't even use character classes (i.e. [ ] )! This is broken regex.

I told you how to make it work. \s for space.

You can use [] as long as they don't contain spaces.
[a-z] is fine for example.
[a-z\s] too.

It's a workaround that makes it unnecessarily difficult to compose and parse when it comes to more complicated strings.

And when you advertise the local search to be a regex search, it's very confusing and frustrating for users when they see that it's *not* proper regex. It's not unreasonable to ask for a switch that enables newsbin to deliver precisely what is claimed.