<div class="gmail_quote">On 15 April 2012 08:56, Nick Sabalausky <span dir="ltr"><<a href="mailto:SeeWebsiteToContactMe@semitwist.com">SeeWebsiteToContactMe@semitwist.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
"Manu" <<a href="mailto:turkeyman@gmail.com">turkeyman@gmail.com</a>> wrote in message<br>
news:mailman.1717.1334400919.4860.digitalmars-d@puremagic.com...<br>
<div class="im">><br>
> Indeed, but I also want to produce some log output. A dialog box is nice<br>
> too, so artists can choose to 'ignore' and programmers can 'debug'.<br>
><br>
<br>
</div>I keep flip-flopping on this.<br>
<br>
At first, I was going to reply and say "No, allowing 'ignore' is bad because<br>
it just means continuing to work within a corrupted state. Only bad things<br>
can come from that."<br>
<br>
Then I realized, "If the artist had a bunch of unsaved work when the error<br>
occurred, it's possible the state might still be good enough for them to<br>
save their work. No point in denying them that for the sake of purity. If it<br>
results in a corrupted save, well...they *are* using version control, aren't<br>
they?"<br>
<br>
But my current thought now is: Shouldn't there just be some sort of<br>
journaling or auto-save or some such? That would eliminate the need, and<br>
temptation, to work through a corrupted state.<br>
<br>
No doubt I'll flip-flop again...<br></blockquote><div> </div><div>The reason more often than not, is that an assert may affect some individual system fatally, and that system will fail, but the artist simply doesn't care about that thing (it's most likely experimental and not ready for them to use anyway). It's usually good practise for the programmer to write code in such a way that a colossal failure in his new experimental system doesn't bring the whole game down if it crashes. The artist just wants to run around and see that his new work is correct when put in context in-game.</div>
<div>So they are usually not DOING work (that they risk losing) with the crashed (but survived) game engine, they are typically just testing their work on context. So if the prior assert's invalid state starts creeping and does bring down the whole thing, they won't lose any work, just a little time an patience :)</div>
<div><br></div><div>It's not unusual to see artists persisting with a buggy build of the game, booting, clicking through 20 assert messages, then finally to a point where it stops complaining and they can see their work, potentially while half the world is broken around them :)</div>
<div>It's not worth wasting a programmers time in this case, so they don't.</div></div>