Talk:Rosetta Code/Run examples

Output extra credit

Since it's just extra credit I don't think this should keep it from being a full task, but I'm concerned about the output checking EC. It might just end up being too problematic because of the different ways languages show certain things. The first example I thought of was lists. Here's how the same list might look in different languages:

Scheme/LISP(s):

(1 2 3 4 5)

Java:

[1, 2, 3, 4, 5]

J:

1 2 3 4 5

Prolog:

[1,2,3,4,5]

All of those are the same list, but they look pretty different. I'm also a little worried about someone taking a solution that does this output checking and using it to mark examples incorrect, but that doesn't seem very likely at all so I'm not that worried--only a little bit. Like I said, it shouldn't keep this from becoming a full task, but people should make sure they think ahead a lot if they try to do this EC. --Mwn3d 17:51, 22 November 2011 (UTC)

yes, that is very true, and the more reason why this is extra credit.--eMBee 18:02, 22 November 2011 (UTC)

Some points

I do have some points for consideration:

1. Layout of examples: examples have not been created to be auto-run and massaging a wide selection of examples to run could make for too long a program. You might consider something like "Assume that code inside the first <lang> tag below the languages {{header|}} tag is all that need run".

   that sounds like a good idea. in my mind i expected each <lang> tag to be a separate solution, so the downloading could would have to pick one or run all of them on by one.--eMBee 18:51, 22 November 2011 (UTC)

2. Would this lead to high server load when developing and testing a solution - especially for the extra credit part of the task?

   Running every example's going to be impractical anyway. Some require extra command line arguments. Some require user interaction. Running a single example for a single language would be enough (where they are nominated at runtime; no hard-coding!) and wouldn't cause problems with excessive server load. –Donal Fellows 15:38, 23 November 2011 (UTC)

3. You might want to just have language A download and run examples from language A. Rosetta Code normally allows a task to be fulfilled in one language without necessarily knowing another to any degree, (if at all).

--Paddy3118 18:36, 22 November 2011 (UTC)

well, the knowledge needed for some of the languages is minimal, eg: to run a Python solution, only python task.py is needed. surely anyone can implement support for that without knowing python. the main point here is to write the program in a way so that support for additional languages can be added easily because the framework is already there. and once more solutions appear, taking them from one language and porting them to another should be easy.--eMBee 18:51, 22 November 2011 (UTC)