Fair enough. I am merely expressing what I believe users would want from a 
shell. Shells without these features feel almost unusable. Perhaps competition 
will suffice in the end.

Glyn
On 12 May 2010, at 15:07, Peter Kriens wrote:

> Tab completion and history are outside the spec ... I do not think it is 
> clever to standardize those, let implementations compete.
> 
> Kind regards,
> 
>       Peter Kriens
> 
> On 12 mei 2010, at 13:05, Derek Baum wrote:
> 
>> On 12 May 2010 03:48, Glyn Normington <[email protected]> wrote:
>> 
>>> I'm sorry I don't already know if these are already provided, but two
>>> critical requirements for the OSGi shell are:
>>> 
>>> 1. tab completion of command syntax and parameters (such as bundle symbolic
>>> names and version numbers)
>>> 2. "up arrow" command recall (preferably filtered by any current input
>>> including that provided by tab completion)
>>> 
>>> Glyn
>>> 
>> 
>> 
>> Hi Glyn,
>> 
>> RFC-147 defines an OSGi CommandProcessor service containing a single method:
>> 
>>   CommandSession createSession(InputStream in, PrintStream out,
>> PrintStream err);
>> 
>> and the primary method in the CommandSession interface is:
>> 
>>   Object execute(CharSequence commandline) throws Exception;
>> 
>> The above defines the RFC-147 "runtime" bundle, which does not contain any
>> mechanism for user input.
>> 
>> Typically, a separate shell or console bundle, not defined in RFC-147, will
>> get input from the user and pass it to CommandSession.execute(). It is thus
>> these bundles, and not the RFC-147 runtime, that is responsible for creating
>> command completion and history etc.
>> 
>> Derek
> 

Reply via email to