Re: [Po4a-dev]Changing the option names? Really?

[Denis Barbier]

On Sat, Feb 01, 2003 at 09:32:36PM +0100, Martin Quinson wrote:
> Hello,

Hi Martin,

> I am about changing the option names of all scripts, but I would like to
> check first what do you guys think. 
> 
> Before the questions, the proposal. I plan to use those options in
> po4a-gettextize, po4a-updatepo, po4a-translate, and maybe in po4a-normalize:
> 
>  -m,--master : original document
>  -M,--master-charset
>  -t,--trans : translated document [I'm not happy with this one. Looking for
>                                    a better name]

-l, --localized?

>  -T,--trans-charset
>  -p,--po : po file
>  [po charset don't exists since the charset information should be clear]
>  
>  -a,--addendum:
>  -A,--addendum-charset
> 
>  -f,--format : module po4a (formerlly -t)
>  -F,--help-format : Lists the available formats

IMO there is no need for a short option here, you may need this -F flag
in the future.

>  -o,--option : I guess we will encounter cases where we *must* pass options
>                to modules. I don't like this [too error prone], but i'll
>                have to.

The GNU way is with --, maybe you could use the same here.
Moreover -o is often a synonym of --output, so I do not like this short
option flag.

>  -d,--debug
>  -v,--verbose
>  -V,--version
> 
>  [po4a-translate specific]
>  -k,--keep: 
> 
> 
> And now, the question: Are you all happy with those names ?
> Is this a problem that the same option is used once for an input
> file, and in another program for an output file (like po is for
> -gettextize and -translate)?
>
> The user will hate me if he calls the wrong binary and if the
> file get lost because of this. I mean, if I unify the option
> names between programs, I've no way to check and say "hey, I'm
> the program which erase the po file and build a new one, not the
> one doing the contrary", like it is for now since the command
> line passed to -gettextize will be rejected by other programs.
> 
> Sharing the exact same syntax between program argument (even
> when [or while] having a different semantic for them) is an
> advantage or a problem?

Do we need all these command line flags for all scripts?  If
scripts are run as filters (i.e. po4a-translate < input > output)
it is much less error prone.  Command line flags could be
provided, but then users are supposed to know what they do.

Denis