Close on decision regarding 0-based and 1-based coordinates in API

https://github.com/xtermjs/xterm.js/pull/2470 added the 1-based interface IViewportCellPosition (uses col/row, aligns with how escape sequences work), https://github.com/xtermjs/xterm.js/issues/2480 will add IBufferCellPosition (x/y) which may be 0-based? Is that confusing? What should we do?

@mofux tells me monaco uses 1-based and language servers use 0-based. Maybe we should just use 0-based everywhere (should we avoid col/row in the API)?

[Tyriar]

Ok, let’s go with 0-based and have the end range x be exclusive so it’s consistent everywhere as I think we have to keep selection like that.

[jerch]

Yes, imho all the confusion comes from natural language usage (or trying to apply simple counting rules - a mistake found in those very early programming languages as well). The 1-based cat is already out of the bag - hard to put it back…

Edit: Or to put it another way - for me the difference is whether we think in indices or item/object counts. Natural languages and/or common logic would prefer to refer to objects (where a 0th thingy does not make any sense). Index usage is very unpopular in such non programming or math related context (maybe we simply dont like indirections if we can just point to the “thing”, idk). Other point is the meaning of 0th here - its just nothing, or null? or empty set? or a non-object that if its the only element in a set would still leave the set empty? If treated in object realms as done in common logic we step right into all the set theory problems - which was not solved/satisfyied until Russell. All in all - “common every day logic” has prolly solved this for ages - nope, go away, thats “nothing”.

On the other hand in computing almost every data layout has an origin and uses indices to drive in some direction from that origin - in this context it is logical to call the origin 0. Most (un)famous prolly being a pointer address itself as offset from the memory segment start (byte index).

For 0th col - isnt that just applying the index to the item itself (kinda trying to skip the origin+index offset indirection)? Within a programming context or formal languages I am fine with this either, but not for common everyday logic. It would just drive ppl crazy and work against common counting rules - “if there is one thing, its the first thing” - “What the 0th thing here?” - Not defined in common context imho (or the universe might rip apart 😈)