Row-wise combine of Coe objects

[MikePapaWhisky]

Any word on the implementation (as hinted in ?combine) of row-wise, or preferably choice between row-wise and column-wise, combining of Coe objects? Would be useful when calculating mshapes (which must be done on Coe) are then combining mshapes with other shapes.

_E.g. might expect the following to re-create bot_coe_

bot_coe <- bot %>% efourier(nb.h = 2, norm = F)
bot_coe_whi <- bot_coe %>% Momocs::filter(type == "whisky")
bot_coe_bee <- bot_coe %>% Momocs::filter(type == "beer")
Momocs::combine(bot_coe_whi, bot_coe_bee)```

[vbonhomme]

soon, eg before the release of MomX, within a month I'd say

[vbonhomme]

any suggestion for self speaking names ? possibly not including c or combine ?

[MikePapaWhisky]

Guess it depends on whether 'combine row-wise' and 'combine column-wise' are going to be two separate functions, or argument options of a single function. Assuming the former...

I haven't found any use for 'combine column-wise' (not to suggest I don't think there are perfectly worthwhile ones), so its tricky to think of a succinct name for it.

For 'combine row-wise' the most obvious suggestion for me would be merge. Otherwise, and in Tidyverse parlance, bind_rows would be the nearest comparison. It would be handy if differences in $fac between the two objects being merged could be handled, making it something like dplyr::full_join. Maybe join?

[vbonhomme]

• combine column-wise is typically used when 2+ views are combined into the same Coe.
• bind_rows/cols would be nice but they would only make sense when one have a very good understanding of what's going on (which is desirable but not the rule). merge is already used in base R.

Based on [this], what do you think of:

• tie for row-wise (it carries the idea of binding two long stuff together isn't ?)
• couple for column-wise (same with the idea of joining similar stuff together but more side-wise just like a couple in real life?)

what do your English ears/brain/heart think of that?

[MikePapaWhisky]

(Yeah, silly of me to forget use of column-wise combine).

Both are essentially 'to concatenate', though not sure which direction more appropriate for that word.

Like tie, but stitch might be better ('tie' has a plethora of meanings in English). Also, though, maybe append (as you're appending data onto data).

couple is good too but might imply the two are kept separate. Other ideas hinge and unite/unify (both of imply bringing together similar things).

Discussed around office, append and unite preferred.

[vbonhomme]

append is also a R base function but sounds good. I think this could be just c (which is short for concatenate as well).

unite is nice but the contrary operation already exist in Momocs (dissolve). gather is already taken, what about assemble?

otherwise what do you think of superimpose and juxtapose. It's perfectly understood by frog people but do you guys use it a lot too? Pretty long though.

---

side comment, this conversation is a pretty good confirmation of this

[MikePapaWhisky]

juxtapose sounds good (you could have appose too) for the combine column-wise function - i.e. to place side-by-side.

'superimpose' not so good as means to place on top of each other, and these functions are for joining things together.

assemble doesn't feel right to me (lacks direction). How about its synonyms convene, accumulate or even `bunch'.

Whatever you think is best, frog ears are better than beef ears anyway.

[vbonhomme]

I think I'll let combine for column-wise and create a c method for row-wise

[vbonhomme]

Is now, natively, solved in Momocs2