(spent my time between previous and this post detailing & checking again that these are the most narrow set of parameters & complete set of starting conditions (dont want to fire of issue i know i dd not test with everything in my ability))

ok, so what does work

in id tags going to an album to the tracklist, select the compilation field with the mouse and typing "1"

regretfully also found several cases where it doesn't work that i couldn't determine what was going wrong.



testing prep:
closed id tags & cleared cache
newly ripped album using perfectmeta data that was correct on first try, compilation was ticked after detecting, ripped accurate, artwork added correctly, default rip profile
album showed as <various artists>, filesystem layout was created as various aritsts\name of the album
metadata via metaflac: tag album artist is not present, and then Compilation= (vorbls_comment comment[9]: Compilation=1)
made a copy after the first rip so i could retest if i was unsure
closed everything, cleared cache, accuraterip show 1 album
changed but likely not relevant: filesystem not alliowed charecter -> i change all result characters to underscore, and added a lot of more to get changed to _ (perhaps should just use force ansi)
closed id tags & cleared cache
i tested with close , clearing cache restarting - as well as just do everything in one session. this made no differnce
as such to keep it daable the following is can be reproduced by using 

these are all in the id tags files & folders + tag edit mode when editing an album.
most of these have subtle differences, so to to keep it understandable these will all start with the following parameters:​


==
what are you testing again?
 -> all these tests are ways i was unable to get compilation=1 set on a track, either by adding as new tag, or updating existing.

 -> --reset-- means set compilation=1 again on all tracks (can be via metaflac, found out how with id tags as wellà


get ready for a bunch of subtle changes :)


right click single track, remove compilation tag -> ok -> right click same track, add compilation via "add" button, it's added to the list with popout box "yes/no", tried these options followed by clicking ok.
click yes, click no, click yes then now, select different, ... actually tested almost everything (this is about when i realized that most of the testing 
result: no compilation tag

this is about when i realized that i was looking for a complex pattern, instead of a simple one: if you do not get an input but (so where you can freely type) dbpoweramp will refuse to even add the compilation tag, let alone set it to "1".


....let's continue with all the tests



--reset--
right click track, delete compilation, ok. album mode select track again, main album now says "Compilation "no"", set it to yes
tag not added

select file without tag and 1 with tag, album now shows <different> for compilation as a dropdown box, options: yes/no/<different>
no option will add tag to track without compilation, but this is where we can start getting ourselves in trouble


--reset--
delete tag from 1 track, select tagless, 2 tracks above and 1 track below, right click, in the change compilation from <different> to yes, ok
now you find yourself with all tracks saying compilation=1 __except__ the tagless now says "yes"


i tested a lot more scenariosn, but as soon as i got myself into a situation where i had some saying 1 others saying "yes" and i started going over the tests i had already done but with 1 track with compilation=yes in the mix there were to many variables.




and than a few more things that threw me off


--reset--

delete compilation from 5 tracks in in sequence, click refresh, click on first empty compilation input field, then start typing 1 & down arrow using only the keyboard until they are all set to one, click refresh, several will remain blank


analysis

i'll stop here and wait for feedback, but i still stand by my original analysis of type confusion. i think it must be clear to you as being the devs that compilation is at times used as a boolean and at other times as a free input field. while i'm not thrilled with that in the end this is a design choice. i do feel my issue is valid since it seems i can select both types in a single action, while often if a boolean and free input select are selected at the same time the boolean is offered as the only input, this can be bypassed.

if i set compilation=a on a track (stupid ofcourse, but i's allowed in the gui) ... well, when and how something changed is close to random.



conclusion

i wish to make clear that i am a new user, i read the fora and manual. i have 500gb of flac from 15 years age (eac & plextor scsi) of which more than i wanted did not pass a new accuraterip check.
also, this started a me trying to figure out why one album showed up as 20 albums with 1 track. while this is still a wall of text i most of these things and now i have a more global view i think i understand the mechanics, i've been fiddling with this for several weeks. at the same time i also ran into some other metadata issues and right now i'm just putting all my compilations aside - have enough other stuff to do.

i also recall what triggered me into digging into this: i had already seen that most albums just had compilation left blank & working compilations had compilation=1
after the x'th time accuraterip said all single tracks i thought "just delete the metadata to the bare minimum", deteted compilation tag from all tracks, added it by mouse selecting all files , right click , add compilation tag. and then i had 1 album with compilation=yes. with a bit over 200 albums done & over 1000 albums to go it seemed a good idea to get my metadata right from the first go.

i have several completely seperate things i was looking into of which several i understand why they happen.

i read the first reply to my question a few hours after it was posted. while i think the original post had enough info i also read through most of these fora and a few others to know i could expect short answers. i also knew the same questions seemed to come back often enough that often times they try the first few google hits & then give up. i spent my time from the previous post till now to test & write this, i know it's a lot to read, but i hope it answers most of the followup question in advance, as well as all the (perhaps uncommon, but with interesting results) ways the i tried to look at the same issue.

danke.





fix

i do not know what framework or housestyle you use, so have no idea of the impact, but the safest & quickest option could be to just force compilation to only have 3 states: true/false/not present & an option to map true/false to either yes/no or 1/0. from what i've read that will cover most needs. you can always leave and advanced option to make it free input and let people shoot themselves in the foot :-D