Re: On computerese

[Findlay Driver]

As making everybody equally unhappy is easier than making them 
happy, I suggest sole use of 'con/catenate'./s

On Sun, Sep 15 2024, Damian McGuckin wrote:
> Somebody on the internet tried to assert that "concatenate" was 
> a
> computer term.

Interesting. I would assert that 'concatenate' is (now) a 
computing term. Personally, I don't remember when I last asked to 
concatenate my order with one consisting of a tea and slice of 
cake. :) I am interested if anyone knows how concatenate became 
the standard term for it's use in computing; it seems like an 
obscure word to reach for regardless of the prefix.

> One comes from the Latin 'catenat' and the other to 
> 'con-catenat-'.
> So, even its usage in Latin is technically separate. Any Latin
> scholars here?
> ...
> I thought that "catentat-" means to "link in a chain" (as in by 
> things
> like atoms or molecules that do it for themselves) and that
> "concatenat-" means to "link together in a chain" by some 
> external
> party, i.e.
>
>         atoms can do the chaining for themselves
> but
>         files need Doug or Holger or myself to do the chaining

N.B. Not a scholar, but I agree.

The 'con-' prefix creates similar subtle distinctions elsewhere. 
> From the etymology of 'connect', I think you could 'nect' (tie) an 
individual something to another (or many into pairs), but to tie 
many somethings into one I would 'con-nect' them. But, as always, 
beware the pitfalls of relating formal Latin to modern English - I 
deliberately use a hypothetical example.

Taking a cue from physics and maths, a 'catenary' (the only other 
time I can think of the root in current usage) refers to the curve 
formed by a rope/string/chain hung from two points. I think you 
could therefore argue that catenate refers more to the link itself 
than the resultant structure. 'Atoms can catenate to others' - we 
are interested in their links. 'A polymer is made of concatenated 
monomers' - we are interested in the 'all together' bit.

Thus, I think that you could argue that when linking many things 
into something 'new' you are concatenating, with a single link 
sufficiently expressed by catenate. When we use terms applying to 
'1, 2, or many', we often choose the more plural for general use, 
hence 'concatenate' and probably partly why I don't use nect in my 
day-to-day. I would argue that cat(1) concatenates files onto 
standard output because the end result of all the successive 
catenations is the desirable part.

I think that there is a similar situation with join/conjoin. I 
program in Clojure, which uses `conj` to add elements to a 
datastructure in the logical/efficient way. 'Conjoin' has a better 
sense of 'joining with the existing stuff', i.e. `conj` returns 
the same type as the input datastructure. Using `join` for this 
operation would not be confusing but suggests a strictly binary 
function with equally-typed arguments, at least to me.

> I have to admit that 40+ years of using "concatenate" would take 
> a lot
> of unlearning.

Of course, `conj` is probably named just so to align with `cons` 
(short for construct (v.)). Convention and familiarity rule here, 
or else use of con/catenate would not be consistent. Why not 
'chain', 'agglomerate', or 'serialize'? I would personally 
consider concatenate an unpaired word with a cranberry morpheme(?) 
and therefore best used in full most of the time, even for a 
grammar-heavy ecosystem as Groff has.

> I am very much on Doug's side there. Isn't "use" being used here 
> in
> its more general use?

Took me a few goes to parse this. I wonder how people feel about 
'usage'?

Thanks to anybody bothering with this for your work on Groff, by 
the way. I've used it to write coursework, essays, and other 
documents since secondary school with all sorts of odd formatting. 
I hope it isn't rude to chime in as a list-lurker, and that I'm 
doing it right; I don't mean to just add to the noise.

Cheers,
Findlay