Strolling through the history of lichess-org/chess-openings on github

Strolling the history of https://github.com/lichess-org/chess-openings

https://docs.github.com/en/search-github/getting-started-with-searching-on-github/about-searching-on-github
understanding-the-search-syntax

https://github.com/lichess-org/chess-openings/commits/master/a.tsv
https://github.com/lichess-org/chess-openings/blob/b7964bd71b7722209181e5cacdec4c4c1053e3ff/a.tsv

https://github.com/lichess-org/chess-openings/commits/master/a.tsv?before=40e56cb7cb885f19405dff9c331bd6f2e00617b0+35
https://github.com/lichess-org/chess-openings/commits/master/a.tsv?after=40e56cb7cb885f19405dff9c331bd6f2e00617b0+34
https://github.com/lichess-org/chess-openings/commits/master/a.tsv?after=40e56cb7cb885f19405dff9c331bd6f2e00617b0+69
split from raw to 5 .tsv
https://github.com/lichess-org/chess-openings/commits/b7964bd71b7722209181e5cacdec4c4c1053e3ff/raw/raw.txt
https://github.com/lichess-org/chess-openings/blob/b7964bd71b7722209181e5cacdec4c4c1053e3ff/raw/raw.txt

blob versus commits.. jargon again..
newest raw? (when split)
https://github.com/lichess-org/chess-openings/commits/b7964bd71b7722209181e5cacdec4c4c1053e3ff/raw/raw.txt
oldest raw 4 years.
https://github.com/lichess-org/chess-openings/commits/ca0f1f4170fa2b24c18b6e775121b020ff798d95/raw.txt?

something about sequence before name (which I support.... the fact is the sequence not the name). to read.
https://github.com/lichess-org/chess-openings/commit/d4ca926895719ea2d73bb7b643e509c88d540af4

7 years ago.. the full history.. raw was 4 years.. I am still lost in Github. word salad:
commit, requests, accepted, completed, verified, merged, closed.
also over 7 years have such pointers changed object pointees?
so. from friend, not from rational navigation from top down from repo.
https://github.com/lichess-org/chess-openings/commits/master
https://github.com/lichess-org/chess-openings/commits/master?after=40e56cb7cb885f19405dff9c331bd6f2e00617b0+454

I would not even be sure that this whole page without any green "verified" stamp were actual changes.
https://github.com/lichess-org/chess-openings/tree/077b028866ce5b654f2f57988ddd126fc9367837

I don't have a complete dev high level model of what can happen even for such a limited repository.. with no executable code.. being just a data repo. with lots of curation imbedded.

inducing as usual. because the forward dictionary then full sentence path.. is unnerving, I would assume in my putative model of GitHub mechanics of changes restricted to this repo.. that any change, before the word commit even is spoken (there are no other branches for testing, but there are many forks, I can't look there), should I assume that all the changes have had a PR made? I guess so.. is merged keyword enough to make a master change

or is that one of the intermediate steps in the sequence flow from birth of PR to master a.tsv file change?

from chess and chess knowledge discussion point of view, my priority in this, to induce from this evolution, I would want such a vetting process.. from issues attached, to acceptance.. but I don't want the infrastructure distractions, I want the chess logic discussion that makes whoever has vetting power decide or accept..

there is the github mystery keywords, and then what github does not stipulate.. it can't even be executable is released or functional. or not a regressions..

at each and every step of the chess discussion that is "surnageante" it would be full of chess theory/knowledge/whatevers stances undergoing some uncovering friction, for my ignorant eyes to latch on...

so first the Github fog.. then the politics of the repo.. current guess.. the is a main contributor..

i have no clue of the vetting system there. how many of those seven years worth of history.. I should look at first..

dropping this.. might learn from examples, most recent as usual.. ahh.. I thought I could download something about changes indicating some patterns of meaning intended by mere existence of sequence in the database..

I know there are factors of popularity.. but that might also be informative.. I have notice the front page has some reasonable guidelines about naming.. I think it is rather arbitrary to demand one name of a position to also fit sequence naming.. not same category of objects.. so restricting criteria such as shortest path.. or most popular path at time of naming (or could be live.. most popular is such reference game database).. or both criteria might solve all the naming fog.. knowledge of path taken for a input position can superseded the pure position input case, which would use mentioned uniquifying criteria. (all questions, most of time, tired of sentence shaping for that.. obvious).

but I care about the sequences themselves. and the act of naming more than the names themselves..

I also would like to make sure or study the "stability" of the principles about what is an opening from a automatic pipeline computer constraint compatible point of view.

how many sequences of positions intersect (I would have to grind all sequences for all positions). There is also the the position ambient space to all this of course. I am still within that spiral of related hypotheses.. nagging me for their lack of dissonance with my slowly evolving chess knowledge so far..

coud my naïve but persistent planar tree mathematical mind's eye model, made of branches with segments, forming various lines be sufficient to generate the zoology of named variants (again not the names of the named, but what is being named out of chess wilderness).. same question we have with PS, where things are more visible perceptible with board logic. The paths taken being in tow. and increasing with more people of high level sharing their knowledge on top of predecessors in that directions (compare wiki page from say 3 or 4 years ago to today, is the job finished).

so segment. positions, any mathematical transposition, and naming determinants is that whole maximal set of move séquences and explicit positions visited... and the names as input data for this segmentation questions..

since babbling is mostly for me.. and seems not to be readable by others.. i should seriously get back to R programming set up.. at home.. with my current reduced stamina.. gambling about can I do it?

prevalence of first transpositions and would a segmentation allow to make a non-turn by turn abstraction (yet function) of the set of all included sequences? there are first transposition positions, and then continuation

the position concept of transposition already extended in the opening repertoire serious students to opening transpositions, could it also be extended to segment continuations..

segment may just be my word for existing currently used after previous discussion about going to fundamental meaning of what is an opening variation.. (before giving it a name or having been named but not about which name as first concern).. there are various ways to go.

It is possible that segment might be having friction with initial standard position being always the start of a game..

that the segments determining the distinction between opening sequence (with definition that it always starts from initial position, not like the segment which would have the branching point as boundary), are not what people mean by variation. but it is what I think is the variation..

sounds silly.. but that need clarification.. might be obvious to the already knowing.. but not to me. mathematically either would work as long as we are explicit about our words and our things pointed by the words..

So, I have high hope about the entrails of the github repository, as it seems to have accelerated in the past years. and seems to be using same structure of repo and file names. I even though it was a recent idea of lichess..

did I link to the very first 7 years old version that seemed to have been about using the ECO source.

the file is ECO.json.. some 700k.. ECO.json..

in other words.. maybe an opening database is not well formed if a segmentation representing all the exactly determined branching points and branch to their deepest positions included, is not also sufficient for the naming problem completeness and distinguishing need..

all segments with branching points or initial position or "leaf" of the set of all full sequences (the "lines").

I guess transpositoin positions, could form their own categories of segments interacting with this first level pass.

that first pass would be the single game single decisoin tree path point of view assuming all nodes (as only described by the move made, not the transition pair of positions) are uniquely defining the line object (and it would, because it seems that the knowledge is not just about the position as ID, but full sequence to position as ID, and I bet all the full sequences in the database are indeed unique).

however, chess being a game of position information in its most elemental description, transposition, which requre at least the existence of 2 game instances in the set (necessary, can't be a unique game, having a transposition, can it?).

so.. second pass.. becoming aware of the ambient space.. grinding all the positions of the game dataset.

another abstraction structure, adding the new informatino about many games sharing some segments subsets.

then discussion of the "topology" of those with respect to others.. it might my words, but I think it might help seeing, should i repeat, seeing, the naming problem clearly.. or have a method to scann the database for some level of consistency and completenes. natural language .

I guess since I am working from mind's eye that my words a failing.. I should make simple trees.. in drawings.

wihtout any SAN distraction.. using thick points for positions when they are explicit in the drawn model, and thin points, for moves.. so much work and I no real digital penmanship.. the overhead procedural steps and search of right tools to the task . make me say.. well that is all I can do.. At least I can think it.

fleeting. i thought I saw another reason against segmentation acceptability.

tired. trying.. before quitting.
I did the full sequence assumption problem (my from position naïve perspective). check
I did mention that transposition question (which should be independent of the naming spag, or prior to), check.

but another.. gone.. at least this trace or deja vu.. memory lapse collision with previous idea being written not seen as same thing.. mysteries of being the observer and actor, and only one brain for both.

I did make an analogy with biological sequences that did not satisfy my explanation intent then, edited out.

but the thought made its way back as I was trying to get away from here.. and re-center my thoughts back to my morning plans about chess study plan, or just let the steam die down.. but sequel afterthoughts.. sometimes with wrapping value.

so. in few words, expressive for me, as just reference to existing lexicon in my memory, so more like a cheat sheet writing.. one day I might refine the analogy I was building (for there are many similitudes but the nuances and caveat to also invoke would make my rambling reputation not budge an iota, too late i started my thought with unlucky sentence beginning, and rewinding is everybody's worse nightmare trap).

so (scroll bar show itself in the input box on the right, danger-danger, arm waving in dislocated directions).

this 2 pass segmentation idea, one for divergence and one for convergence, kind of reminds me of conservation versus convergence in evolution. It had fooled many a meticulous database curator. But in chess we have been heavy on the conservation point of view historically. So the challenge might be to study convergence. And I think named PSs is a good position information structuring pair of googles to look at opening monster, which then might have some stars to work with for position based navigation. load of analogy in one sentence.. tapping own shoulder.

we have named divergence everywhere in terms of paths difference. named PSs are looking from another angle.

in the practical problem of making a framework of full sequences to completely cover the lichess userbase wilderness in their whole game database, and have some discerning (separating the important signals), I find that forgetting about transpositions first with the usual non-chalant assumption that hand-move makes a move, would recover the conservation (genealogy tracking emphasis), and then working on a more position information complete and discerning superimposable view would allow us to complete the decision tree view to account for the board being there before any decision sequence would have to be pondered..

it might be how I can illustrate the need for an acknowledgment of the éléphantesque ambient position space.

ok. I might be assuming that the segment do not change the size going forward...
or more mess.. but I think the segmentation before any naming purpose might be more rational way of characterizing the input opening database as set of full opening seqeuence. what did jomega use temporarily the master sequence? was it to allow my segmentation idea on top? it seems compatible.

but for me segmentation can happen independently of the naming of variations .. I am just hoping that they would coincide with the manifest of emergent pure move sequence branching points, in ignorance of any naming, by virtue of presence or inclusion in the opening database or framework as jomega so wisely made his set of studies (older repo source basis).

the idea of using branching points is just a natural consequence of having many seqeuences.. or one could say the set of master sequences define a tree of branchin points (or decisions) from which a corresponding segmentation would be a mathematicla consequence.. then overlay the naming propositions.

I suspect that is not how things are done.. more about one pair of name and sequence being introduced and its ripples with existing names.. but I may be wrong and someone is actually checking the whole set of sequences for all its possible emergent branching and possibly even checking for all positions (as EPD without opcodes) happening in many of the sequences of the database as variable (enumeration definition of database is the variable input to branching skeleton abstraction, which I conflate with segmentation, as all branching points, leafs and roots would be boundaries to their adjacent segments.. if line is already accepted why not segments, and branching point obvious scissor points. (enzyme restriction.. in delete paragraphs).

I guess. I just have to merge all the sequences in a spreadsheet and learn my R programming way through some wrapper with python chess eventually.. for grinding sequence set into positoin set.. and use the banskia idea of SQL based relational database. to not have to fit restricted chess expectations of chess specific software. with either expensive proprietary, or olded GUI... (still better than some atom or electron based browser mimicking a GUI).

but since I prefer to spend my little energy left for learning ware stuff.. I would rather learn to deal with SQL.

or some graphdatabase extension of that.. I kind of recall some libreoffice database standard that was compatible with I think neo4j.. graphdatabase.. big thing....

first merge and grind.. find a way to hold big positoin database (well big for me)..