The Old New Thing

One hour later and no comments yet? Raymond, your pre-emptive comments are a killjoy.

Back in my IBM days doing OS/2 there was "slime" when things were shutting down for a release.  The release management team would prevent things from going in that appeared to be risky, but an open defect regarding a spelling error, or other very-safe-sounding-to-fix bug would more than likely be allowed into the build.

Groups would keep a trivial defect open and not fix it so that they always had a way to sneak big changes into the product late if they felt they had to.  It's my understanding that OS/2's entire plug-n-play infrastructure, "device manager"ish control panel, and updating all the standard device drivers was slimed into the build right before golden-master time under the guise of a "Sound Blaster device driver doesn't detect IRQ correctly on a white-box PC" defect.

"Pre-emptive Igor Levicki comment"

ROFL, thank you

Pre-emptive stupid comment: "Chicken sucks."

At my office we generally get pools going on those sorts of things.  Basically anything that could conceivably be gambled on most likely will be.

In our group we call 'remora' checkins 'limpet bugs'. I suppose the intent is to compare to mollusks, but 'limpet' makes me think of 'limpet mines' -- magnetic naval mines used to sink ships. perhaps apt.

The snarky comments were funny - especially the last one.

JamesNT

You make it sound as if products ship bug-free-ish.  From experience with being involved in beta programs, there is another dev team behaviour which is along the lines of "must close out as many bugs as possible so a) our dev team doesn't look as bad as others, and b) total product bug limit is below what we can ship product with".

"...but 'limpet' makes me think of 'limpet mines'...

I never knew that - 'limpet' only made me think of Don Knotts. Child of the media, ya' know.

I believe Iain McDonald introduced the "shark" and "limpet fish" terminology for bugs during the Windows 2000 end game. Maybe it's more of an Aussie usage.

You know, this is why Microsoft products suck.

[igor]Especially Vista.[/igor]

We have this end-game period called "LBF", which stands for Limited Bug Fixes.

In every meeting and announcement, I refer to that as "Lots of Botched Fixes".

My peers want to kill me :P

Actually, the sane reason to avoid being the last one in as the final moments are when enough changes can happen that your fix can break the build. And blame for breaking the build always goes to the "winner" who has to prove that his checkin didn't break it.

Fun came the other day, when the most senior top guy checked in an innoculous change. His new file wasn't used anywhere (yet - he wanted it in code control). I get called over when the build broke (moi? I touched nothing, but I knpw enough to figure out where blame is to be cast.). He accidentally checked in code through a symlink - which the build always recreated as part of the build prep stage. Of course, source control didn't make a symlink, but a real directory... the build script couldn't create the symlink... good thing the script checked return values or it would've been very strange during testing.

Ah, well you should have used a superior system such as CVS that requires you to check in the directory hierarchy first ;-)

Or SVN, which doesn't dereference symlinks when you check stuff in...  ;-)

There's a similar thing where I work, too.

Personally, I've always commented that it was a bad attitude for development teams to take since I could see a shy and withdrawn developer might keep their mouth shut about something they see just so they don't get teased.

When I bring that up to management, I'm told it's all in good fun, so, I wonder if it really doesn't matter.

Sometimes chicken != developer. It might be management.

Management:"Hey you're all done? Since there are plenty of time left, why don't you add these new features?"

Developer : <reluctantly>"O Kayyy." </reluctantly>

Man Igor is famous now. He's in Raymond's blog!

I bet Igor's site hit skyrocketed.

I like that word, innoculous.  A perfectly cromulent word.

>>Pre-emptive Igor Levicki comment: "Windows Vista sucks."<<

I can't decide whether Raymond likes me or hates me or a bit of both but it seems that I have become an important part of his life when he is devoting his valuable time to pre-empting me.

>>Man Igor is famous now. He's in Raymond's blog! I bet Igor's site hit skyrocketed.<<

I don't even have a counter on my website.

Back on topic -- it seems that there were considerably less chicken involved in making Windows Vista otherwise it wouldn't need SP1 so soon ;-)

PingBack from http://www.levicki.net/articles/blogments/2008/02/28/Last_Checkin_Chicken.php

WinXP was released in 2001, SP1 was released in 2002 IIRC. I don't see how Vista is that different except that it was first released in late 2006.

I must be a masochist, I love being the last commit on a project.  Sometimes I'll commit minor code refactorings or just one more test case to get it.  

Not sure exactly why but it gives me a feeling of getting in the last word and probably some control freak as well because I know when I commited all the tests passed and the build was solid so no one else better commit and screw things up :)

Hearing that others dreaded it was a bit of a surprise.