The difference in gameplay wouldn't seem to be huge. I suppose it could theoretically allow a player to research several levels of a tree in one turn with high research levels
The difference will be at most one turn of extra time to research a tech, which is genearlly insigificant, as most techs will have minimum research times of 5 or more turns. Note that even if a player has enough RP generated per turn to pay a tech's whole cost at once, it still takes the minimum research time to complete, due to limits on how many RP/turn can be put towards a tech. So, you'll never actually be able to research several levels of a tech in one turn, as you'll aways run up against the per-turn limit on a multi-turn tech. The multiple levels in one turn scenario is only relevant for testing cases where techs cost 1 RP for 1 turn, in which case you could pay the whole cost on one turn, if allowed.
Enqueued techs that are un-researchable should be drawn in similar colors to the ones in the tree, to indicate they are un-researchable.
Already happens, both on research and production screens (the latter if you have a build item enqueued that can't actually be produced this turn for some reason).
Marking techs that are normally un-researchable, but will get some spill-over research is somewhat problematic. I can't think of an obvious way to distinguish them, but if we need the feature, something could be devised that will make sense to the player if he reads the help.
It's probably not that important. Both on the queue and on the tree, colouration indicates whether a tech is researchable immediately, which is defined as having its prereqs already researched. How long it will actually take to be researched is indicated by the number of turns to completion, which are reasonably accurately predicted regardless of whether the tech is currently researchable.
It would be nice if queueing a tech automatically queued all dependencies before it.
That's the plan. Before I could do that though, I needed to have the queue handle enqueued techs that aren't yet researchable, during which the present issue came up.
I am presently pondering what order in which to enqueue the prereqs though. I can traverse the tree backwards, and add all the techs to the queue easily. But presumably it'd be nice if they were in order with the earlist prereqs first, then the later ones... and perhaps ordered by lowest to highest cost after that, but I'm unsure of how specifically to decide between ambiguities.
You may have to differentiate between explicitly queued techs and techs queued to satisfy deps. This would be for the case where a dep is X or Y. The line to get X might not be known when you've explicitly queued a tech and so the line leading to Y gets queued, but at some later point tech X may be discovered/stolen/bought and so the explicit tech should be researched and the techs automatically queued to satisfy deps should be dropped.
This likely won't be an issue, as I believe we'll make it impossible to steal or buy a tech for which you don't already have prereqs, and make it take several turns to do the stealing or to finish learning the tech you bought. This prevents gameplay exploits (such as what Moriarty seems to like doing), and is more realistic (contrary to his claims).
As for techs enqueued because you enqueued something that depends on them, but then removed the tech that depended on them, I'd rather just leave the extra techs on the queue. It's simpler and less error prone, particularly since all the prereq techs have other uses than just being prereqs for the tech you initially enqueued, and it's not necessarily the case, after removing the originally enqueued tech, that the player will no longer want any of its prereqs.
Moriarty's post was based on faulty assumptions. See reply to eleazar's post above: either way you can never research many levels of techs in one turn, except for test cases that are unrepresentative of the actual game's techs. All techs have minimum research times, presently with numbers like 5, 10 or 20 turns, though these will change with balancing.
I don't see why he/you calls it more realistic to research many levels at once though... You need to spend time learning the basics before the more advanced stuff can be learned. That doesn't happen immediately, regardless of how good your lab equipment is or how much money you throw at the researchers / students.