Thread: Spelling

A have a small note on the exciting subject of orthography.  I have
developed a DocBook-aware spell-checking toolchain, which keys off the
markup to decide which parts of a file contain English text, as opposed to
program listings, etc.  Since we have not applied markup in super-pedantic
ways (and don't worry, I don't plan to police you), the signal-to-noise
ratio isn't exactly pretty right now, but the test has still turned up a
fair number of typos, honest mistakes, and inconsistencies.

If we can migrate to a consistent spelling in the documentation and the
programs themselves, it would give a more pleasant feeling to the system,
IMHO.  The following is my view of the rules -- feel free to correct me.

* "multi" is not a word by itself, so there is no hyphenation when it's
  used as a prefix:

  multibyte character, multicharacter operator name, multiversion
  concurrency control, multicolumn index

* Many terms that are written as one word when used as identifiers in
  computer programs should be written as two words in English, such as:

  file name, user name, host name, data type, file system, time zone,
  time stamp, index scan, bug fix, query tree, range table

  When these words are used as adjectives then they are hyphenated, e.g.:

  time-zone database, bug-fix release; also: built-in function

* Spelling unclear:

  runtime -- Probably better "run time" (cf. build time), but it looks weird.
  netmask -- I'd prefer "net mask" (cf. subnet mask), but "netmask" seems
             to have established itself.
  oid/Oid/OID -- probably "OID"
  timeout (noun) -- Dictionary has "time-out", which seems right.
  mergejoin -- probably "merge join"
  mergejoinable -- probably "merge-joinable"

--
Peter Eisentraut   peter_e@gmx.net   http://funkturm.homeip.net/~peter

...
> * "multi" is not a word by itself, so there is no hyphenation when it's
>   used as a prefix:
>   multibyte character, multicharacter operator name, multiversion
>   concurrency control, multicolumn index

No arguments here, but...

I *think* that it is typical and accepted to use hyphenation in cases
where the base word does not have that qualification as common usage.

An example might be (but then again, may not be ;) "multi-version" vs
"multiversion". There is "a version", and there are "multiple versions",
but in common usage the multi prefix is not associated with this noun
(and what the heck, the phrase is using it as an adjective anyway!). So
hyphenation is acknowledgment that the word is constructed in a way that
is different from the usual practice.

As a counter-example (ooh, there's a hyphen now ;), the opposite of
"multiversion concurrency control" may be "single-version concurrency
control", with the hyphen required (substituting a space is not right,
and running the two words together isn't either). Hmm, but maybe the
opposite is actually "universion concurrency control"??

Oh, it's too confusing. But I'm on the record as liking hyphens.

                    - Thomas

> ...
> > * "multi" is not a word by itself, so there is no hyphenation when it's
> >   used as a prefix:
> >   multibyte character, multicharacter operator name, multiversion
> >   concurrency control, multicolumn index
>
> No arguments here, but...
>
> I *think* that it is typical and accepted to use hyphenation in cases
> where the base word does not have that qualification as common usage.

I agree with Thomas.  The other changes Peter suggested are fine.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026