How can you keep customers and developers separate from each-other?

Why would you want to do this?

Does the solution scale to multiple products?

How would you keep different customers from seeing other customers' bugs?

or similarly w/ html:

What is the real problem?

What is the right fix?

If necessary to get this fix into bugzilla proper, you can make the default template not do that and switch your template to do it. then you contribute your changes, and argue in favor of switching by default. And yes, I do endorse this change. There's a bug for this of course but don't ask me for its number.

The engineer asks me for the number... Anyway, the fix shouldn't be hard, (the engineer agrees it's trivial), and it is so much more useful than a custom hack.

The engineer observes "bzbot is gonna need some smacking around because it uses LWP to get data.. and therefore needs an account." My off the cuff response is that "bzbot already runs into that problem", well sorta, on rare occasions. Really when bzbot encounters a classified bug, mozbot2 has trouble playing with it.

The engineer debates just putting a hook in BugMail to ping the bot if it exists.

Why does bzbot need lwp at all?

The engineer responded "[to] get [the] product" I responded that his fix is silly.

It is, and he agreed, but why? read on

• Support Read Only access to profile data
• Allow merging/reconciling of profile data (Bookmarks, History, Cookies)
• Inactive profile users should flush their changes and relinquish their lock(s)
• There should be a single universally recognizable lockfile
• Instead of tossing up a dialog and refusing to allow the user to use the profile data, we should let the user use the profile and start a thread to watch the profile lock/data. If the lock disappears we should acquire it and offer to reconcile the profile data
• Consider splitting the locks

What can be done to catch stale locks?

• We can ping the computer to see if it's alive
• For local processes, we can check to see if the process actually has the lock file open

What can not be done to the profile lock?

• The global lock can't be deleted. Deleting it would result in a big backwards compatibility problem: an old browser (like Netscape 6) would think that the profile is unlocked and ignore the new split locks.

Oddly enough just about every language you can imagine has some measure of a statement. Even English. Some statements may be simpler than others, but the number of statements is still a better estimate of complexity than the number of lines or bytes.

So how would one count statements?

For XML, I would count each (nonwhitespace) node that would be generated in a dom.

This means that <foo></foo> and <foo/> would both have value "1", and <foo>bar</foo> and <foo/><py q="q" r="r" s="s" t="t"/> would both have value "2".

For JavaScript and Perl a statement is generally delimited by a newline or a semicolon.

For C/C++, statements are generally delimited by semicolons or parentheses.

x=x+1;++y;r*=5+3/p; /*3 statements*/
for (a=0;b<4;c++) b+=c; /*5 statements*/
while
(
0
)
{
}
/*1 statement */

An icon containing 4 images would be assigned a value "4". And an xbm containing 1 images would be assigned a value "1". This means your average xbm loses value (sorry).

I'm puzzling over derivative works.

Suppose we have an interface for "nsIScreenScraper" and an implementation for it on windows. Suppose someone makes a new implementation of nsIScreenScraper on Mac OS. The implementation is functionally equivalent but is in a sense a translation. The original implementation is MPL (for argument's sake). Would you claim that, by virtue of the original implementation being MPL, the Macintosh implementation must be MPL or receive permission from the author to be something else?