Nicolas Goaziou <n.goaziou@gmail.com> writes:

>> 2. HTML symbols have been tested in Firefox.  In a few cases I
>>    couldn't find a pretty name (like "&pi;") in which case I've
>>    supplied the unicode number (like "&960;").  Is that OK?  (E.g. can
>>    Org produce non-uft8 HTML?)
>
> I think it can: see `org-html-coding-system'. It may be wiser to avoid
> these symbols altogether.

OK.  The only ones remaining are 

    ("smiley" "\\smiley{}" nil "&#9786;" ":-)" ":-)" "☺")
    ("blacksmile" "\\blacksmiley{}" nil "&#9787;" ":-)" ":-)" "☻")
    ("sad" "\\frownie{}" nil "&#9785;" ":-(" ":-(" "☹")


> Any additional package needed must be added to
> `org-latex-default-packages-alist', which means that the package will be
> loaded in every LaTeX export. So, this is clearly not OK for packages
> like "kpfonts".

OK, they should all work.

>> I also fixed some
>>    inconsistencies.  E.g. loz was loaded diamond even though loz is
>>    defined in amssymb and "~" and "\tilde" produced different results
>>    (the latter produces the irritating "\~{}" often observed among
>>    (social science?) university professors...).
>
> Sounds good.

>> + ("preccurlyeq" "\\preccurlyeq" t "&prcue;" "[precedes or equal]"
>> "[precedes or equal]" "≼")
>
> This one is LaTeX specific, it may be more appropriate in
> `org-entities-user'.

There are many X-specific entities. . .  E.g. lArr is html specific.
One important feature of entities is that unicode symbols can be
displayed in buffer.  If you feel strongly about it, I can remove
them.

–Rasmus