DRAFT Whitepaper: Dealing with Obsolete error messages

re: DRAFT Whitepapr: Dealing with Obsolete error messages

Stuart Ballard — Mon, 22 Nov 2004 18:27:00 GMT

Shame that #pragma doesn't work in VS2003 :(

re: DRAFT Whitepapr: Dealing with Obsolete error messages

Neyah — Mon, 22 Nov 2004 19:03:00 GMT

Nice article. It looks to me like after reading this whitepaper, you couldn't really have any other questions as to how to handle the obsolete warnings. You explained the problem, gave example problems, and example solutions.

Couple of nit-pick typos :
Header has Whitepapr instead of Whitepaper.
Under Summary 2), #pargma instead of #pragma.

One that might confuse people:
Where discussing XmlSchemaSet vs XmlSchemaCollection, you put :
... you will discover that XmlSchemaSet was obsoleted ...
rather than :
... you will discover that XmlSchemaCollection was obsoleted ...

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Brad Abrams[MSFT] — Tue, 23 Nov 2004 04:44:00 GMT

Thanks Neyah!

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Brian McManus — Wed, 24 Nov 2004 00:11:00 GMT

I've been trying to do exactly the same kind of improvement to our .net codebase using the Obsolete attribute, however I was reluctant to do so because we use "TreatErrorsAsWarnings".

We also use FXCop so this approach is perfect; except that the current FXCop (1.312) does not contain the rule ConsiderNotUsingObsoleteFunctionality!

I haven't had a chance to check my VS2005 with FXCop built in to see if there are any additional rules [also seems to be a bug with running it straight after build] so I'd be interested to know if there will be an FXCop update any time soon?

Regards,

Brian

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Brad Abrams [MSFT] — Wed, 24 Nov 2004 00:27:00 GMT

Thanks for your comments Brian… your reluctance around TreatErrorsAsWarnings is exactly the issue I hoped to address with this paper. Please let me know if you think our guidance there is clear enough. I hope it makes your job easier.

On the FxCop rule… well, that is another reason why this is just a draft, the rule is not yet ready for prime time. It is not in any public FxCop drop, but the FxCop guys assure me they will have it for Whidbey RTM.

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Brian McManus — Wed, 24 Nov 2004 09:32:00 GMT

The guidance is spot on Brad.

The only other comment I would make is regarding developers using this attribute.

While I understand the article is primarily guidance around framework generated messages it may be useful to point out that this attribute can be used to promote evolutionary improvement in any .net codebase/components without a big bang effect.

Keep up the great posts,

Brian

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Brian McManus — Wed, 24 Nov 2004 12:35:00 GMT

.... or you could just link to my post on the subject :-)

http://seattlesurfer.blogspot.com/2004/11/evolution.html

Brian

1.0 || 1.1

Dela's Ramblings — Fri, 26 Nov 2004 04:13:00 GMT

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Michael Giagnocavo — Fri, 26 Nov 2004 17:38:00 GMT

I think it's nice to give guidance on how to disable this warning. I suppose the fault doesn't lie with the BCL, but the compilers that generate warnings.

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Joshua Flanagan — Tue, 30 Nov 2004 19:05:00 GMT

I would like one clarification:

"When this is the intent the member will be marked with Error=True in the ObsoleteAttribute indicating that a compiler error (rather than warning) should be generated. You are strongly recommended to migrate your projects off these members to ensure they work seamlessly on future versions of the .NET Framework."

You say "strongly recommended to migrate"... but does the developer really have a choice? If Error=True causes a compiler error, I would think you would HAVE to stop using the obsolete type/member.

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Brad Abrams [MSFT] — Tue, 30 Nov 2004 19:22:00 GMT

Yup, Joshua, that is a fair point... you do have to move off those Error=True members if you move to the new version. I will update the paper.

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Stuart Ballard — Tue, 30 Nov 2004 21:48:00 GMT

I hope/assume there will be some Intellisense magic to ensure that anything marked ObsoleteAttribute(Error=true) will not show up?

Actually, it would be great if anything Obsolete at all showed up in a different color (grey, for example) and had low precedence in the string matching algorithm...

re: DRAFT Whitepaper: Dealing with Obsolete error messages

Brad Abrams[MSFT] — Tue, 30 Nov 2004 22:18:00 GMT

Excellent idea... let me float it buy a few people.

1.0 || 1.1

briandela — Thu, 01 Jun 2006 16:51:48 GMT

Dare Obasanjo aka Carnage4Life - C# 3.0 vs. Ruby: Thoughts on Extension Methods and Open Classes

Wed, 23 Jan 2008 07:00:36 GMT

PingBack from http://www.25hoursaday.com/weblog/2008/01/23/C30VsRubyThoughtsOnExtensionMethodsAndOpenClasses.aspx