IFDB search game for “adventure” used to list games with adventure in title including Crowther & Woods’s Advent. But now it goes to this game directly without listing any others:
5 Likes
I’ve noticed that when I search “Never Gives” it says there are 0 games, even though I have a game with that name. Searching just “never” or “gives” does the same thing. I wonder if the recent restructure of search (that made things faster) causes these issues.
4 Likes
I had the same issue when trying to add NGUHD to the “longest games ever” thread. I success at the end.
1 Like
This has now been reported on github:
1 Like
Thanks for raising these issues. They’re actually due to two different problems.
The “Adventure” issue is caused by our changes to support implicit search by IFID. Christmas Adventure is configured to have a bunch of non-reasonable IFIDs.
Since its IFID is “ADVENTURE,” we’re matching that one directly.
“Never Gives” is due to our our change to use a fulltext search index. Out of the box, our database uses a large list of “stopwords,” words that will be ignored in all searches, but the default list is more than 500 stopwords, words that absolutely shouldn’t be stopwords.
4 Likes
The IFIDs seem to match the tags for the game, so maybe these were entered as IFIDs by mistake.
3 Likes
I marked Dan’s post as the solution. I will edit the title of the topic to indicate as solved when both issues are fixed in IFDB.
It has been in my mind when I created the topic to tag it ifdb, but it totally slipped out of my mind to actually do it, thanks Hanon for doing it for me.
3 Likes
I’ve filed PRs to fix both of these issues. I’ll post an update here when they’ve merged in.
2 Likes
I’m glad these were small, discrete changes rather than major systemic issues!
3 Likes
Should they be removed in that case? Not sure if there’s a rule that IFIDs have to be UUIDs, but I think basically all of them are. Since it’s the easiest way to avoid collisions?
2 Likes
The intent is that they’re all UUIDs except for old games that predate the IFID spec, which get special-case IFIDs computed from the game file contents. As with all indexing schemes, this has not worked perfectly.
But in this case, it seems like a clear metadata error.
4 Likes
In the PR for this issue, I just removed the feature to implicitly search by IFID. You can still search by ifid:adventure and Christmas Adventure would turn right up.
If somebody wants to clean up bogus IFIDs, I think that’d make sense. Here’s how I found them.
Initially, I queried for IFIDs that didn’t contain a hyphen, like this:
select ifid from ifids where ifid not like '%-%';
That turned up 320 IFIDs, most of which looked like this:
| ifid |
|---|
| 004D9E39231A1797D3A8A58ADA7C02F4 |
| 00CC0D46A7A54FD08D89FB4652DA3CDC |
| 0182F9777087D22739413F6B7C1896FA |
| 01A08F7D37CB02940A1D6EAA7E57E5F9 |
| 04FC82FF2020CED57E315493435C59C1 |
Do those need to be “fixed up”? I’m honestly unsure. They appear to be 32-character UUIDs without their hyphens.
So then I searched for all IFIDs where the length was not exactly 32 or 36 characters long, and where the IFID does not start with any of the known, declared prefixes from the Treaty of Babel.
select ifid
from ifids
where
ifid not like 'ADRIFT-%'
and ifid not like 'ADVSYS-%'
and ifid not like 'AGT-%'
and ifid not like 'ALAN-%'
and ifid not like 'GLULX-%'
and ifid not like 'HUGO-%'
and ifid not like 'LEVEL9-%'
and ifid not like 'MAGNETIC-%'
and IFID not like 'MZ-%'
and ifid not like 'TADS2-%'
and ifid not like 'TADS3-%'
and ifid not like 'ZCODE-%'
and length(ifid) not in (32, 36);
Here's the list of 147 plausibly invalid IFIDs
| ifid |
|---|
| 02479785 |
| 1 |
| 110422 |
| 120115 |
| 120324 |
| 120528 |
| 121216 |
| 1212162134 |
| 123423232 |
| 130419 |
| 1313133 |
| 13911 |
| 140826 |
| 150525 |
| 160314 |
| 180104 |
| 181115 |
| 1989 |
| 210814 |
| 22083007 |
| 223423423 |
| 227788 |
| 230409 |
| 252204787 |
| 2A4C8FF-0CB9-451A-85AB-4271D14CAC29 |
| 2HWV41LDJGHB3D4L |
| 3138023 |
| 33113311 |
| 359F3A66ED77FC3D4C1FA2DCB640A447B5E6CB344404F2FBC1DEF8A7A51C6400 |
| 3F866853-BC6C-4927-8F0C-BFFFDF6C3E7 |
| 407E5657-D5E9-4218-BE3E-3DECA658485F. |
| 4406415 |
| 616263 |
| 6623 |
| 69950 |
| 6XBYTLWLC1C1ZT0G |
| 74VDSUAU3KRW9048 |
| 7777333999555 |
| 78173 |
| 7A7865B2-DB4C-9574-48910D475317 |
| 823-8978-8888-8888 |
| 8888880119 |
| 8B0747EF-B9DB-4CA9-94B8-84960E73373 |
| 8CA5115F5-B05A-4070-A79B-0C65E8758AE5 |
| 943EBD54-BBF3-4457-A31B-66FB83991020. |
| 9782384 |
| 9A3CFF0B-4CDF-453B-9CA3-453A28447B5A. |
| A6C2C73A-0358-481B-BFBF-C461BEE04515. |
| ABXKHCI33ZSWEB2L |
| ADVENTURE |
| ANASTASIYA |
| BADGERCIAR |
| BASIC |
| BEKANTDOOM |
| CAMELOTPARADOX |
| CCU403250023 |
| CDF1 |
| CHRISTMAS |
| COM.CESTRIAN.DEEPSAGA.ANDROID |
| CRAWL |
| DATING |
| DUNGEON |
| EUROPOP-VAMPIRE |
| F2E89303-A810-4012B581-BAFB00A61AE7 |
| FANTASY |
| FIDO |
| G@BEN |
| GNOMONROSE |
| HAUNTED |
| HEXFU |
| HITLER |
| HOLIDAYS |
| HOMINATEH |
| HOMINATEHH |
| HOMINATEHHH |
| HORROR |
| HOUSE |
| HPECD |
| HTTPS://INTERACTIVEADVENTURES.ITCH.IO/WHAT-DWELLS-WITHIN |
| IDK |
| IFRIT10169 |
| INSOMNIA |
| INTERACTIVE |
| INTEVENCIONIDENTIDAD |
| JACL-002 |
| JACL-003 |
| JACL-004 |
| JASPSTIC |
| JOURNEY1979SOFTAPE |
| JOURNEYTOLUNARNINE! |
| KG5UM5UZVJZ1VJSL |
| KLUEANDDOO |
| KUMA74262ZH466DL8K |
| LAPDOG |
| LIBRARYSERIALNUMBER991113 |
| MARKCMARINOCOM |
| MATCHMADEINSTEELAARTHUR9 |
| MICRO |
| MIDELLEBLESSINGS |
| NA |
| NARRATIVE |
| NIGHTDRIVE474 |
| OR-1 |
| OR-2 |
| OR-3 |
| OR-4 |
| OR-4E |
| OR-5 |
| OR-6 |
| PARANORMAL |
| PB3SNH6NK4VNS8PD |
| PLZ_NO_PLAY33 |
| REDOCK001 |
| RISEOFTHEMEMELORDS |
| RQ:BRR+ |
| RSGSK1JGLWYNDJH4 |
| SAFESIMFIREV1 |
| SAMFOXALL |
| SANTA |
| SCRIPT-57E69A4362FB89E2E4237D65C8280C67 |
| STORYSMASH |
| SURVIVETHEHOLOCENE |
| TAPLO-1 |
| TAPLO-2 |
| TFLIZZLE |
| TLAOTLCYOA |
| TLHWBB |
| TRANSGRESION |
| TSB4JDMN4FHDPEBH |
| TURRAKS-CLASSICS-THE-PRISON |
| TWINYJAM |
| UHYES |
| UNABLETODETERMINE |
| UNKNOWN |
| UNKNOWNALPHA |
| VWMDIAC |
| WILBURC |
| WIP |
| WK3 |
| WORMO |
| WUT |
| XM9LATSQMA9F6ENS |
| XXXXXXX |
| Y8ZH5-VC1KMSGQ4BXAGEFA |
| YG9X8NZC81JQJTHQ |
| {STXXDI3PYVQ39HN1} |
It’s plausible that someone should just search for all of these (by searching IFDB for ifid:02479785 etc.) and just delete all of those bogus IFIDs.
Finally, as you can see, there are currently no validation rules for IFDB’s IFIDs. Reading over the Treaty, I think someone could cook up a validation algorithm. I’d accept a PR for this, but I’m probably not going to do it myself.
6 Likes
I think I’ll go through and clean up some of these ifids but I’ll start from the bottom of the list in case someone else already started from the top.
3 Likes
If you do try to fix them up, you’ll likely run into this bug that @JTN just told me about:
Edit a game, blank out its IFID field, and save it. You’ll see this error:
An error occurred updating the game record in the database. You might try the request again in a little while, or contact us if the problem persists. If you need to contact us, please tell us exactly what you were trying to do (if you can take a snapshot of the screen showing all of the updates you made, that would help), and mention this error code, which might tell us more about what’s going on internally to the database: IIF070F.0 [1062 - Duplicate entry ‘’ for key ‘PRIMARY’].
I’ve fixed it in a PR, not yet merged. Despite the error message, the change will go through correctly if all you’re doing is blanking out the IFID field and submitting.
That’d be me then… looks like we met around T. (I’ve been cherry-picking the works with obvious bogus word-like IFIDs, leaving the strings-of-letters-and-number ones; and bombing through pretty fast.)
2 Likes
Now these 38 weird IFIDs remain.
| ifid |
|---|
| 1212162134 |
| 123423232 |
| 130419 |
| 1313133 |
| 13911 |
| 150525 |
| 160314 |
| 180104 |
| 181115 |
| 210814 |
| 22083007 |
| 223423423 |
| 227788 |
| 230409 |
| 252204787 |
| 2HWV41LDJGHB3D4L |
| 3138023 |
| 33113311 |
| 4406415 |
| 616263 |
| 6623 |
| 69950 |
| 6XBYTLWLC1C1ZT0G |
| 74VDSUAU3KRW9048 |
| 7777333999555 |
| 78173 |
| 823-8978-8888-8888 |
| 8888880119 |
| 8B0747EF-B9DB-4CA9-94B8-84960E73373 |
| 8CA5115F5-B05A-4070-A79B-0C65E8758AE5 |
| 9782384 |
| ABXKHCI33ZSWEB2L |
| CCU403250023 |
| CDF1 |
| KG5UM5UZVJZ1VJSL |
| KUMA74262ZH466DL8K |
| PB3SNH6NK4VNS8PD |
1 Like
I’ve had another pass through the numbers-and-letters ones, and got rid of some of them.
Most of what I’ve left was works where it’s impossible to tell what the real IFID is if any, because there are no links at all from the IFDB record. (With some vague notion that the bogus IFID might be A Clue, to someone who cared more.)
1 Like
I think this regex allows all valid IFIDs, while not allowing a lot of invalid ones. It will allow a few legacy IDs which actually don’t exist, like LEVEL9-025. It assumes that the language codes for Level 9s Champion of the Raj are two-letter codes in uppercase although this isn’t specified in the treaty.
^([0-9A-F]{8}-([0-9A-F]{4}-){3}[0-9A-F]{12}|ZCODE-\d{1,5}-[A-Za-z0-9-]{6}(-[0-9A-F]{4})?|GLULX-([0-9A-F]{8}|\d{1,5}-[A-Za-z0-9-]{6})-[0-9A-F]{8}|((TADS[23]|HUGO|MAGNETIC|LEVEL9|ADVSYS|ALAN|HTML|MZ|ELF|JAVA|AMIGA|SCRIPT|MACHO|MAC)-)?[0-9A-F]{32}|MAGNETIC-[1-7]|AGT-\d{5}-[0-9A-F]{8}|LEVEL9-0[0-2][0-9](-(\d|[A-Z]{2}))?)$
This should be matched as case sensitive (without the “i” flag).
The regex should be self-explanatory 
2 Likes
This one is in fact valid.
It’s missing the Adrift IFIDs, and the updated legacy Hugo ones. (The Babel spec hasn’t had a formal update since these were documented, so check the source on GitHub.)