Lee Daniel Crocker wrote:
See, that was my point, trying to conform to the dot convention is trying to maintain compatibility for compatibility's sake, without regard to the utility or the situation at hand.Didn't you just tell me that "Too much /really bad/ software has been created by praying to the false God of backward compatibility..."? :-)I don't care about one application either: I care about violating a basic decades-old useful convention of the Unix operating system:
Yes, exactly; what's your point? I'm advocating that we abandon backward compatibility here (that is, the ability of newer versions of the server to read the old dot-named folders) to serve an entirely different purpose: following convention.
You view the dot as a path separator (and therefore a prefix for subfolders of INBOX) as a broken feature of Binc, I view it as a compatible feature with other IMAP servers (specifically the Courier that Binc-Is-Not-), thus enabling Binc to be a drop-in replacement.
If Binc were to eliminate dot prefixed subfolders of INBOX, then Binc will no longer be a replacement for Courier. This leaves three distinct options:
1) Maintain compatibility with Courier IMAP for subfolders of INBOX
2) Break compatibility with Courier IMAP by changing the path separator
3) Do both via some kind of configuration mechanism
I think you're not noticing that .foo.bar is "INBOX.foo.bar", thus it is the separator we are in fact talking about.This is the heart of our difference of opinion, you consider it to be a bug to prepend folders with '.', I don't. I encourage you to write a patch that makes the path separator configurable.
It has nothing to do with the path separator: that can still be ".", and I'm happy. The server just shouldn't create folders named with a ".", and that's a completely separate and unrelated issue to the separator.
And what I'm arguing for here is that the notion of changing something that already works to make it work differently (but as you insist, better), is secondary to making Binc comply with the IMAP protocol which allows for root folders.I would rather time be spent on furthering Binc's IMAP conformance than accomodating other software's interpretation of reality.
That's what I'm arguing for here: conformance. And not just to RFCs, although that's a good starting point. It should also conform to user expectations about the way software works in Unix. That "other software" was here first. It is our job to conform to them, not the other way around.
Changing something that works is not as important as adding something that is missing.
No, the primary attribute of usable software is utility. For me, that comes down to solving more problems than it creates.The primary attribute of usable software is /humility/. The world is littered with the carcasses of "good ideas"; the software that people actually use is the software that plays well with others.
C=)
--
--------------------------------------------------------------------------
Better the hard truth than the comforting fantasy. -- Carl Sagan
--------------------------------------------------------------------------
Caskey <caskey*technocage.com> /// TechnoCage Inc.
--------------------------------------------------------------------------
A presumption on your part does not constitute an obligation on my part.
