hi guys... so what do we do? Patrick says that our patch could be restrictive because we cannot user options for argument like captain would like... And Patrick added that they were some implementation problems...
do we decide to use opthash normal? or do we implement an option -strict which specifies that an option must not be passed as an argument... see ya guys On Nov 20, 2007 7:35 AM, Patrick Davalan <[EMAIL PROTECTED]> wrote: > On Mon, 19 Nov 2007 13:42:15 +0100 > Kryz <[EMAIL PROTECTED]> wrote: > > > > > Hello Patrick, > > > > The logic of the hack is to not allow an option to be put in the > > command line instead of an argument. > > > > Elseopt returns an error when a valid option is typed as an argument. > > > > For now this is done by making an array with the options. > > Opthash check the options and arguments in the command line. > > if there is an arguments we check in the options tab, if its true we > > return an error message else we continue in the loop. > > 1 ) Conceptually it's a bit restrictive, > > You may not type : > $ myscript.rb --file -f > which probably means that you cannot use a file '-f'? > You may want this, but maybe sometimes you won't in some scripts. There > should be a switch to enable/disable this feature at parser creation > and/or some way for the user to override it. > > 2 ) There are some implementation problems. > > - You assume that an option may have 2 names like '--file' and '-f', it > is not exact, an option may have n names ( n > 0 !), for instance > '--file', '-f', '--path', '-p' etc. > This should only be checked by the length of the array which describes > an option. > > - Moreover, opthash uses getoptlong to parse the command line (which, > for some other reasons, may not be a good idea), getoptlong accepts > unambiguous abbreviations for options, that's to say : in a script > which accepts 2 options '--directory' and '--debug', in place of > '--directory' one may enter '--director', '--directo' ... '--di'. > (I'm not sure it is typically a good idea but it's typically boring to > implement). Therefore, to check for an option name given as an argument, > matching this argument against the names given to opthash is not > sufficient. > > I don't know what the urgency of the opthash fix is, it's likely not on > the citical path, thus "il est urgent de tergiverser" and you may first > think of all improvements needed. > I think we should get rid of getoptlong or fix it, it's under the Ruby > licence and there is no legal problem. There some other possible > options like '--' which may stand between options and non-argument > options etc. > > > -- > Patrick > > > -- > [email protected] > Unsubscribe: [EMAIL PROTECTED] > Archives: http://lists.else-projects.org/archives/devel > Jabber-MUC: [EMAIL PROTECTED] > > -- --** alxx **-- George Clooney : what else ? -- [email protected] Unsubscribe: [EMAIL PROTECTED] Archives: http://lists.else-projects.org/archives/devel Jabber-MUC: [EMAIL PROTECTED]
