Re: Translating trace messages

[Paul Eggert]

> From: Akim Demaille <address@hidden>
> Date: 28 Aug 2002 16:38:38 +0200
> 
> 2002-08-12  Paul Eggert  <address@hidden>
> 
> 
>       * lib/bitset_stats.c (bitset_stats_print_1, bitset_stats_print,
>       bitset_stats_read, bitset_stats_write): Wrap strings in _() if
>       they need translation.
>       * src/LR0.c (state_list_append, new_itemsets, get_state,
>       append_states, generate_states): Likewise.
>       * src/assoc.c (assoc_to_string): Likewise.
>       * src/closure.c (print_closure, set_firsts, closure): Likewise.
>       * src/gram.c (grammar_dump): Likewise.
>       * src/injections.c (injections_compute): Likewise.
>       * src/lalr.c (lookaheads_print): Likewise.
>       * src/relation.c (relation_transpose): Likewise.
>       * src/scan-gram.l: Likewise.
>       * src/tables.c (table_grow, pack_vector): Likewise.
> 
> Actually, it was on purpose that they were not marked as to be
> translated: they are not meant for the users, so I don't think the
> translators should have to translate them.

Sorry, I didn't know that --trace was permanently not meant for users.
(I thought that you hadn't gotten around to documenting it.   :-)

If you like, I can install a patch to undo those changes.

> Am I going against some rules?

Not as far as I know.

> Should I have stated this somewhere to make the intention explicit?

Perhaps we should have a new macro that means "This string is not
translated and this was intended?"  Naah, probably overkill.