Forum search ignores search characters

[Michael.Uray]

If I search for
"esp32-c6" ethernet
then the search says it ignores the "C6", but I want to have the C6 included in my search.
What is wrong?

[ESP_Sprite]

Yeah, phpbbs search is... let's say nonproductive at the best of times. Best solution is use e.g. Google with 'site:esp32.com esp32-c6 ethernet'.

[RandomInternetGuy]

@ESPSprite, while I agree that phpbb should have been left in previous centuries this actually IS something that you (the royal "you" at ESP, not necessarily you personally) can control.

Admin->General->Search(down at the bottom) here is a setting for. "Min characters indexed by search". Change this to 2 rather than than the default of 3. (Changing it to "1" builds a pretty ridiculous index to search "a" or "I".) You probably need to delete and recreate the search index. This will probably take a bit of trim and may lock the board for posts while rebuilding it.

It's a worthwhile knob to turn. It'll really help the users of this board as so much relies on suffix-specific chip features. It's a minute of admin time and some amount of compute time that probably ranges in the minutes to tens of minutes range for the machine.

Please do pass this request along to those that admin this site. It'll ultimately reduce for for you and all the awesome ESP_helpers that keep this group.

[ESP_Sprite]

I'll see if I can give it a go, but the issue is that the forum sometimes gets hit hard by spammers / AI crawlers / ..., and I really don't want to blow up a database giving MySQL even more work. FWIW, we're kinda in the process of moving to a new forum software suite (at the moment Flarum and Discourse are candidates) and we hope the search there works a bit more sane.

[RandomInternetGuy]

I can respect that. I've run public forums (including multiple phpbb) and defending them against scum is challenging.

FWIW it was my experience that the only difference of turning this knob down is the SIZE of the index. It increases because it adds an index of the article number (and, I think, offset - it MIGHT do a scan of the field once it's pulled the record from the DB. A 'rep scasb' is fairly cheap once you've taken the DMA and moved it to L1 ) of every two character sequence. On really large forum (and just looking at the article numbers that's a primary incrementing key and having some idea of the valid traffic here, you're not. Sorry.) it makes a difference on article writes. Other than size increase on a post there's very little practical difference. You could put the search page behind login (ISTR there's a plugin for that.), but IMO it's unworth it. If you're hit with a DOS attack and have 100K people executing a search simultaneously, you have other problems.

I don't know the load on your system. I don't know your traffic. I do know this market and specifically this software reasonably well and it's my experience that it won't negatively impact things. It's my experience as a user here (I lurk a lot more than I post.) is that being unable to search terms like "s3" and "c6" is frustrating.

I'd recommend trying it. You have two options (reverting it, password protecting search) that cost minutes each. I think it'll be helpful.

Or, The Group That Names Things could quit naming EVERYTHING ESP32-nn.

I'm done contributing writing to Discord. It's totally unsearchable. The only person you can ever help is the person that asked you the question in realtime. A month later, someone else with the same question has to ask it again. So I don't get a vote but I hope that you and the rest of the team have spent enough time on forums of similar groups to weigh heavily on that.

This post has 66 two-:alnum: tokens.

[ESP_Sprite]

You make a good point, lemme see if I can tweak it and we'll see where it goes.

I'm done contributing writing to Discord. It's totally unsearchable. The only person you can ever help is the person that asked you the question in realtime. A month later, someone else with the same question has to ask it again. So I don't get a vote but I hope that you and the rest of the team have spent enough time on forums of similar groups to weigh heavily on that.

Note we're talking about Discourse, not Discord. Two different things. E.g. the Arduino forums run on Discourse. I 100% agree that Discord would be a horrible thing to base a knowledge base like these forums on.

Edit: I can't find that setting. Possibly the version of phpbb we're running doesn't have it, it's not the most recent iteration ever.

[RandomInternetGuy]

> You make a good point,

We're engineers. We make data-driven decisions.

> Discourse, not Discord

Excellent point. I dislike the Arduino discourse forums, but I don't have a raging tate for them like I do the discord groups.

> I can't find that setting.

I'll admit that my PHPBB experience was, oh, 2005-2013 or so, so it's probable that I'm misremembering the location of the option.

https://www.phpbb.com/community/viewtopic.php?t=1929625
https://www.phpbb.com/community/viewtopic.php?t=1494295
https://www.phpbb.com/community/viewtop ... p=15810861 (looks promising)
https://www.phpbb.com/community/viewtopic.php?t=2571806 is more recent (notice the article number) but it's hard to tell of there's anything actionable there instead of bickering. At the very least, it should be a refresh of what the knobs are called these days.

I'm confident that given that little tweak of motivation you can find more info on it.

Fluffing The Common Word list can help you drop those "a", "an", "I" style words.

Hail from about the era when I would have run into this. They mention details I'd forgotten about like the backend choice. It's pretty comical that the "massive server" (which was too wimpy in 2009) isn't even the size of a decent Pi these days.

ISTR that MySQL changed beween ISAM and InnoDB and full text indexes got inherently better just from being built more for "web era" problems than as a COBOL replacement.

Also, even if you're not REALLY a database dev but have the chops to pull up the tables in a clickable table query viewer, it can be insightful to see what your table sizes and records counts already are. It can also be paralyzing when you see numbers that seem big, but are small to a computer.

My "teach an admin to fish" lesson here is done for the day.