few comments
Ingo
This is a very useful feature! Great work!
I have a few comments/suggestions,
* Maybe it is not only for shim sharing (tasks used as a glue), but
also (scripting) tasks not available as a web service, such as a
script performing a statistical analysis, or a clustering etc.
* Maybe it is a good idea to create a Taverna plugin, which directly
connects to this shim database and that can be used to search for shim
services. These plugin, then, can help the workflow designer to import
these shim services as beanshell tasks. Of course, the script designer
needs to add information about the input and output variables (maybe
through the "script upload interface" or by means of Java Doc or Java
meta data information (@ sign)
* A preview of the script implemented may be useful to decide whether
the script is the one needed (in case documentation is absent).
* I believe syntactic and semantic information about the inputs and
outputs is very important to help the workflow designer to decide how
this shim is essential to decide how he can use the script in his/her
workflow model.
* Better categorisation (and orthogonal). Now, two types are
available: Types and Categories. The distinction between BioMOBY in
types and Moby in Categories may not be clear to everyone. Use terms
like "Task type" for service type, scripting language, and Fuction for
the operation to be performed instead? The names of the "categories"
are not always clear. For example, what does the category "NCBI" mean?
Scripts provided by the NCBI or Beanshells to access NCBI databases?
Good luck!
Ingo Wassink
Mannie Tagarira
Thank you for the feedback you provided. It was really insightful and
is greatly appreciated.
> Hello all,
>
> This is a very useful feature! Great work!
>
> I have a few comments/suggestions,
> * Maybe it is not only for shim sharing (tasks used as a glue), but
> also (scripting) tasks not available as a web service, such as a
> script performing a statistical analysis, or a clustering etc.
Ideally, we would like the Shims Library to use the repository aspect
of myExperiment for script-based shims, and scripts in general. This
would be a great topic to start on the myExperiment-discuss list as we
would need to get the myExperiment team involved as well.
> * Maybe it is a good idea to create a Taverna plugin, which directly
> connects to this shim database and that can be used to search for shim
> services. These plugin, then, can help the workflow designer to import
> these shim services as beanshell tasks. Of course, the script designer
> needs to add information about the input and output variables (maybe
> through the "script upload interface" or by means of Java Doc or Java
> meta data information (@ sign)
I agree, a Taverna plugin would be really useful; we have had some
indication of approval from some Taverna users concerning this. Given
that the project has a foreseeable future, this would be a highly
feasible option.
> * A preview of the script implemented may be useful to decide whether
> the script is the one needed (in case documentation is absent).
We will keep this in mind; this is turning out to be a very popular
request.
> * I believe syntactic and semantic information about the inputs and
> outputs is very important to help the workflow designer to decide how
> this shim is essential to decide how he can use the script in his/her
> workflow model.
We would also like to provide the user with more metadata to help them
understand the shims more. Would you be able to provide a short list
of what you think would be really useful attributes to include?
> * Better categorisation (and orthogonal). Now, two types are
> available: Types and Categories. The distinction between BioMOBY in
> types and Moby in Categories may not be clear to everyone. Use terms
> like "Task type" for service type, scripting language, and Fuction for
> the operation to be performed instead? The names of the "categories"
> are not always clear. For example, what does the category "NCBI" mean?
> Scripts provided by the NCBI or Beanshells to access NCBI databases?
Scripting language info would be great, seen as a shim can be in any
language. Thanks!
I agree, the categories are not spot on; with your help we would like
to provide categorisation that is more appropriate. Could you jot
down a few categories under which shims could fall?
I do not quite understand what you mean by terms u suggested,
particularly "task type" for service type, and "function" for the
operation performed. Would you be able to expand a bit further please?
Regards,
Mannie