validation of result score

[garemoko]

Learninglocker does not accept statements with an empty score raw, score max or score min, but these should be accepted. See #106

Interestingly, it does accept statements where the score values are in quotes. These are rejected by Wax but accepted by Watershed (http://tincanapi.com/public-lrs/). Watershed then returns the statement without quotes. I assume that Wax is technically correct, (I wouldn't expect anything less from @fugu13) but I'd prefer to follow Watershed on this one and accept the statements, correcting the syntax as they come in. Learning Locker currently returns them with quotes, making it incompatible with Wax.

I guess TinCanJS should be validating this, but I also need to update my statement crafter. cc @brianjmiller

[fugu13]

Hi Andrew,

Indeed, you've guessed right :) .

I'd strongly recommend properly validating them as numbers. I can guarantee
you no LRS will be considered conformant if it accepts numbers as strings:
https://github.com/adlnet/xAPI-Spec/blob/master/xAPI.md#lrs-requirements-3

"The LRS MUST reject Statements

• with strings where numbers are required, even if those strings contain
  numbers."

There's good reason for it, too. The only way the standard is going to
realize a vision where it is easy to move data around and switch/use
multiple LRSs is if validation is strict. That way, if data works when sent
to one LRS, it will work when sent to another LRS. If conformance isn't
strictly defined, then we'll enter the interoperability morass SCORM did.

A data model standard isn't really a standard if it isn't exact.

Sincerely,

Russell

On Mon, Mar 10, 2014 at 6:33 AM, Andrew Downes notifications@github.comwrote:

Learninglocker does not accept statements with an empty score raw, score
max or score min, but these should be accepted. See #106#106

Interestingly, it does accept statements where the score values are in
quotes. These are rejected by Wax but accepted by Watershed (
http://tincanapi.com/public-lrs/). Watershed then returns the statement
without quotes. I assume that Wax is technically correct, (I wouldn't
expect anything less from @fugu13 https://github.com/fugu13) but I'd
prefer to follow Watershed on this one and accept the statements,
correcting the syntax as they come in. Learning Locker currently returns
them with quotes, making it incompatible with Wax.

I guess TinCanJS should be validating this, but I also need to update my
statement crafter. cc @brianjmiller https://github.com/brianjmiller

Reply to this email directly or view it on GitHubhttps://github.com//issues/113
.

[garemoko]

I stand corrected; we do need to reject those statements. 

Sent from Samsung Mobile

-------- Original message --------
From: fugu13 notifications@github.com
Date: 10/03/2014 17:28 (GMT+00:00)
To: LearningLocker/learninglocker learninglocker@noreply.github.com
Cc: Andrew Downes mrdownes@hotmail.com
Subject: Re: [learninglocker] validation of result score (#113)

Hi Andrew,

Indeed, you've guessed right :) .

I'd strongly recommend properly validating them as numbers. I can guarantee
you no LRS will be considered conformant if it accepts numbers as strings:
https://github.com/adlnet/xAPI-Spec/blob/master/xAPI.md#lrs-requirements-3

"The LRS MUST reject Statements

• with strings where numbers are required, even if those strings contain
  numbers."

There's good reason for it, too. The only way the standard is going to
realize a vision where it is easy to move data around and switch/use
multiple LRSs is if validation is strict. That way, if data works when sent
to one LRS, it will work when sent to another LRS. If conformance isn't
strictly defined, then we'll enter the interoperability morass SCORM did.

A data model standard isn't really a standard if it isn't exact.

Sincerely,

Russell

On Mon, Mar 10, 2014 at 6:33 AM, Andrew Downes notifications@github.comwrote:

Learninglocker does not accept statements with an empty score raw, score
max or score min, but these should be accepted. See #106#106

Interestingly, it does accept statements where the score values are in
quotes. These are rejected by Wax but accepted by Watershed (
http://tincanapi.com/public-lrs/). Watershed then returns the statement
without quotes. I assume that Wax is technically correct, (I wouldn't
expect anything less from @fugu13 https://github.com/fugu13) but I'd
prefer to follow Watershed on this one and accept the statements,
correcting the syntax as they come in. Learning Locker currently returns
them with quotes, making it incompatible with Wax.

I guess TinCanJS should be validating this, but I also need to update my
statement crafter. cc @brianjmiller https://github.com/brianjmiller

Reply to this email directly or view it on GitHubhttps://github.com//issues/113
.

—
Reply to this email directly or view it on GitHub.

[fugu13]

I'll give you three guesses who made sure that language was in there, and the first two don't count...