Wildcard search working as intended? [opgelost]

13 berichten • 138 keer bekeken

Dit bericht gaat over: rapporteer een bug/fout

Status opgelost
Stemmen voor: 0
Stemmen tegen: 0

» Snelle toegang tot het laatste bericht

HI all, I was just wondering if the wildcard search is working as intended - I have prepared a small but hopefully descriptive test scenario for this:

 

As an example database entry I took this Dutch Duit: N#88859

 

For an easy test we set the “Gregorian date” filter to: 1617

 

And now the important part;

 

Advanced search > “Lettering”: TRA IEC TVM > successful search

Advanced search > “Lettering”: TRA*IEC*TVM > unsuccessful search

 

Is this because of the paragraphs inbetween (that's what I suspect…)?

If so: Is that intended? Depending on how a database entry's lettering is formatted, this might seem like a hard to handle hurdle when trying to identify a coin by its lettering.

As far as I am aware, the wild card only functions with one asterix and it has to be at the end. So it searches term* + anything possible added to 'term'.

 

on*

→on, one, once, onyx, ongoing, oneself, …

Just tested it and can't acknowledge your assumption (would have been an easy solution, unfortunately it's not the case).

 

My test: Searching for this coin N#84237 by…

 

Setting “Gregorian date” to: 1780

 

Setting “Lettering” to: *4 EINEN THAL* > successful 

Setting “Lettering” to: 24 EINEN THAL > successful

Setting “Lettering” to:  *5 EINEN THAL* > not successful

Setting “Lettering” to:  *24*NEN*HAL* > successful

 

In this case the wildcards work as intended. Paragraphs seem to interrupt the logic.

Hello,

I made a fix so that the wildcard * works with multiline legends.

 

Status gewijzigd naar opgelost (Xavier, 22-okt-2024, 15:42)

Thanks for the fast fix! Unfortunately this seems to have broken something within the search option itself. Short test:

 

This is the database entry I want to find: N#3425

 

Base data:

 

Issuer = Germany

Face value = 10

Gregorian date = 1949

 

Test cases:

 

Lettering = …

 

*B*D* > successful (expected)

*B*D > not successful (expected)

*B*R > successful (expected)

*BA*R* > successful (expected)

*BANK*DEUTSCHER* > successful (expected)

 

This shows that the lettering search works in general.

If you now remove the base data mentioned above and only search for the lettering (e.g. *BANK*DEUTSCHER*) - so without any additional parameters - you will run into a database error:

 

The database encountered a problem. Please try to refresh the page in a few minutes (press F5). (MySQL 23-Oct-2024 09:23:03)

 

Is it because this search causes some kind of overflow? Sometimes I might want to search for lettering only, so this is a crucial function for me. This definitely worked fine before the new design launched.

Thank you for the follow up.

 

Searching with lettering with wildcards is very slow in database. I made an optimization to speed up the search. I tested that the example “*BANK*DEUTSCHER*” doesn't produce a timeout error anymore.

Hello,

It seems that the optimization is not sufficient in all cases. I was searching for “*L*NVS” without any other criteria, and it failed with the above mentioned message (The database encountered a problem. Please try to refresh the page …).

 

I understand that it might be difficult to optimize the search with the wildcards lettering criterion alone. Thus I tried the same request by adding a weight range and it worked. 

Do you recommend this technique of combining with other criteria (weight, size, …)  ? Will it lower the search load in any case ? If yes, this might be added in the tooltip of this criteria to help the user going around this issue.

Just tried it with no problem. 

This is not the same search. I was speaking of the “lettering” filter, which builds a different kind of SQL request:

 

 

The intention being to search in a specific field.

 I tried a simple test just now in the Lettering section 

 

CHARLES = 3709 results found 

CH*RLES = 3712 results found [for some reason]

*H*RLES = error message 

 

 So why should an extra * mess a search up like that? 

 

Token collector [1600-1899] with some coins

 Wondering why 3 extra results, I tried 

CHERLES / CHIRLES / CHORLES / CHURLES 

 and all gave no result. Mysterious. 😵‍💫 

Token collector [1600-1899] with some coins

There are probably some ‘Arles’ etc. besides the 'Charles'.

ZacUK

 Wondering why 3 extra results, I tried 

CHERLES / CHIRLES / CHORLES / CHURLES 

 and all gave no result. Mysterious. 😵‍💫 

chemin arles

chemins arles

also show up :-)

Just call me Bram

No new swaps for the moment, still too many half-ongoing swaps to clean up!

» Forumbeleid

Gebruikte tijdzone is UCT+2:00.
Huidige tijd is 16:54.