[Opensim-dev] Error detection when storing an asset

Mon Apr 21 16:05:02 UTC 2014

See below:

From: opensim-dev-bounces at opensimulator.org [mailto:opensim-dev-bounces at opensimulator.org] On Behalf Of Jim Williams
Sent: Monday, April 21, 2014 11:57 AM
To: opensim-dev at opensimulator.org
Subject: Re: [Opensim-dev] Error detection when storing an asset

No!  The lower level code should return what happened, and there is nothing which is "exceptional" except some children; which is the right way to view the notion of exceptions -- dim-witted.  Given that the stupid idea was invented, I must admit that sometimes there is no real choice but to accept the crappy code and propagate the crap -- but there is no reason to do so willingly.  The try/catch scheme is bad practice.  How to deal with the crap that is out there is a matter which can be discussed, but not with the notion that is was done right to begin with.

Then design the api to return success/failure as a return result.  Point is, you don’t choose to gleefully ignore error details from a lower level.  I get that you don’t like exceptions.  And that there is long standing religious debates on how they should be used.  My point wasn’t so much about the mechanism as the approach.  Lower level code returns success/failure and upper level code decides what to do.

It goes along with thinking that objects are somehow special.  Took me years to figure out that objects were simply good programming practice with a bit of crap tossed in to confuse people.  (BTW-- I worked at Object Design.  I know what a C++ object is better than you do.)

I wouldn’t necessarily make that assumption.  You don’t know what I’ve done in my career.   I was a long time user of Object Design though and am well aware of the issues it had with exceptions unwinding the stack.   Just because C++ exception handling sucked for many years though doesn’t necessarily mean exceptions==bad.

Mike

P.S. As Hiro pointed out, I’m not a core member just an interested techie with an opinion who is trying to share it.  

On Mon, Apr 21, 2014 at 11:13 AM, Mike Chase <mike.chase at alternatemetaverse.com <mailto:mike.chase at alternatemetaverse.com> > wrote:

That’s a shame but I understand.  I don’t understand the resistance to what I guess I consider good programming practice because there seems to be concern about “fixing” code that might be effected.  IMO lower level code should throw exceptions if something exceptional happens and let the upper levels decide to retry or report the error.

Mike

From: opensim-dev-bounces at opensimulator.org <mailto:opensim-dev-bounces at opensimulator.org>  [mailto:opensim-dev-bounces at opensimulator.org <mailto:opensim-dev-bounces at opensimulator.org> ] On Behalf Of Oren Hurvitz
Sent: Monday, April 21, 2014 10:54 AM
To: opensim-dev at opensimulator.org <mailto:opensim-dev at opensimulator.org> 
Subject: Re: [Opensim-dev] Error detection when storing an asset

James, the only philosophical concern that I've heard is a desire to hide errors and present a false facade of 100% success to users. This was obfuscated by an attempt to claim that the errors could someday be fixed automatically, but that doesn't actually happen in today's OpenSim so it's not a valid point. Since I respect my users and value their time I have to let them know when I've failed to complete an action that they requested, so I reject this approach.

A second possible objection is a practical one: will propagating exceptions cause other parts of the code, unrelated to the assets service, to fail? I don't know the answer to that, but I still refuse to pretend success in cases where I've failed.

For these reasons, I have changed Kitely to propagate exceptions thrown in the communications system; it is already done. If this causes problems then I will see them and fix them. But in view of the surprising and vigorous opposition to this change I will not be contributing it to OpenSim.

Oren

On Mon, Apr 21, 2014 at 5:43 PM, James Stallings II <james.stallings at gmail.com <mailto:james.stallings at gmail.com> > wrote:

Parhaps it would be interesting to hear whether Oren has obtained to some elegant way of addressing these concerns.

Cheers

On Sun, Apr 20, 2014 at 11:08 PM, Trinity <trinity93 at gmail.com <mailto:trinity93 at gmail.com> > wrote:

Im in mels camp on this one.

Trinity Bays

On Sat, Apr 19, 2014 at 4:14 AM, Melanie <melanie at t-data.com <mailto:melanie at t-data.com> > wrote:

These "other places" are what I'm worried about. There are a lot of them and each of them would need to have code added. Exception handling code is one of the worst types of code because the "try" is a scope, so locals devlared in the try, like bool result = MethodToTry(); will have to be split up into declaring the bool outside the scope and assigning it inside - incredibly ugly for code that wants to be reference and teaching code as well as a functioning system.

- Melanie

On 19 Apr 2014, at 08:02, Oren Hurvitz <orenh at kitely.com <mailto:orenh at kitely.com> > wrote:

My fix has two parts.

The first is that the Store() operation needs to understand failures correctly. There has been a consensus that it should, so I'll add that to Git.

The second is that MakeRequest() should propagate exceptions, instead of just returning null (which is what it does now). So far there have been 3 votes for this (me, Mike Chase, and Justin) and 2 against (Melanie, Diva). That's very close; does anyone else want to make their position known?

Next, I see that there's confusion in this discussion about what happens in Store() if MakeRequest() throws an exception. And the answer is, nothing will be different, because Store() already correctly catches exceptions. That is precisely how it should work: the low-level communications system reports when it has failed, and higher levels (that know the business value of the call) decide how to handle it. However, MakeRequest() is called from other places as well and they might need to be changed to handle exceptions better.

_______________________________________________
Opensim-dev mailing list
Opensim-dev at opensimulator.org <mailto:Opensim-dev at opensimulator.org> 
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev
_______________________________________________
Opensim-dev mailing list
Opensim-dev at opensimulator.org <mailto:Opensim-dev at opensimulator.org> 
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev

_______________________________________________
Opensim-dev mailing list
Opensim-dev at opensimulator.org <mailto:Opensim-dev at opensimulator.org> 
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev

-- 

===================================

http://osgrid.org/
http://simhost.com
http://twitter.com/jstallings2

_______________________________________________
Opensim-dev mailing list
Opensim-dev at opensimulator.org <mailto:Opensim-dev at opensimulator.org> 
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev

-- 

Oren Hurvitz
VP R&D
Kitely Ltd.

Email: orenh at kitely.com <mailto:orenh at kitely.com> 

_______________________________________________
Opensim-dev mailing list
Opensim-dev at opensimulator.org <mailto:Opensim-dev at opensimulator.org> 
http://opensimulator.org/cgi-bin/mailman/listinfo/opensim-dev

-- 
No essence.  No permanence.  No perfection.  Only action. 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://opensimulator.org/pipermail/opensim-dev/attachments/20140421/ee58f6e9/attachment-0001.html>