[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: Fwd: RE: [cgmo-webcgm] More thoughts on exceptions
I don't have really strong feelings about the exceptions. As a rule, I like fairly strict error handling, but in this case I'm satisfied to with a reasonable tradeoff between implementation effort and error detection/throwing. I do have a couple questions and comments. Question. What I think you are saying for Style Properties (and similar stuff) is this: a.) exception for non-existent property (set or get). b.) but no exception for trying to set out-of-range values. Correct? (That seems reasonable to me.) One other comment in-line. Otherwise, I would support a consensus proposal between Stuart and Benoit, since they seem to have converged. Will someone write such a proposal as editing directives and give it to me? (After approval by TC.) Thanks, -Lofton. At 09:09 AM 9/6/2005 -0400, Benoit Bezaire wrote: >This is a forwarded message >From: Galt, Stuart A <stuart.a.galt@boeing.com> >To: "Benoit Bezaire" <benoit@itedo.com> >Date: Friday, September 2, 2005, 9:43:59 AM >Subject: [cgmo-webcgm] More thoughts on exceptions > >===8<==============Original message text=============== >Benoit, > >I just got my computer exchanged at work and am trying to put the pieces >together to make my computer functional again... > >My goal was to mostly raise questions, and look for inconsistancies. I >agree that putting in an exception might not be worth the cost of >implementation. >I agree that making a note in the spec about what happens with bad input >is the better solution. > >Stuart Galt. > >-- >Stuart Galt >SGML Resource Group >stuart.a.galt@boeing.com >(206) 544-9925 > >-----Original Message----- >From: Benoit Bezaire [mailto:benoit@itedo.com] >Sent: Friday, September 02, 2005 5:28 AM >To: cgmo-webcgm@lists.oasis-open.org >Subject: [cgmo-webcgm] More thoughts on exceptions > >I have some implementation feedback to share with the group... Here are >the APIs that throw exceptions (based on Stuart's document). > >setAttributeNS() - throws INVALID_CHARACTER_ERR: Raised if the specified >qualified name contains an illegal character. > >Problem: What is an illegal character? Is it check that they are part of >[2] Char: >http://www.w3.org/TR/2004/REC-xml11-20040204/#charsets >or is it something else? Since we are referring to the namespaceURI parameter... In that case, I think we're talking about 3.1.1.3+3.1.1.4 restrictions, right? I.e., it has to be a valid URI for WebCGM (incl. any fragment bits, following 3.1.1.3 rules), and has to use the ASCII+ repertoire (rfc3986?), with the URI escaping mechanism (3.1.1.4) for non-ASCII+ characters. > - throws NO_MODIFICATION_ALLOWED_ERR: Raised if this >node is readonly. >Problem: When do we have a readonly node where a namespace attribute >cannot be set upon it? > > >setStyleProperty() - throws no exception > >Problem: Stuart asked if we should throw an exception if the user types >in an invalid style or value. I agreed at first, but now I'm having >second thoughts. It seems like a lot of work for very little benefit. >The spec should instead say that out of range values are clamped to >acceptable values. > >setAppStructureAttr() - throws no exception > >Problem: Stuart made a similar argument that was made for >setStyleProperty. I agree that we should throw an exception if a user >was trying to set a new apsid; but I don't think we should look into >invalid values (again way too much work for little benefit). > >removeAppStructureAttr() - throws NO_MODIFICATION_ALLOWED_ERR No >problem: I'm agreeing to Stuart's proposal. > >Thoughts? > >-- > Benoit mailto:benoit@itedo.com > > >===8<===========End of original message text=========== > >FYI, > >-- >Best regards, > Benoit mailto:benoit@itedo.com
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]