On 03:31 pm, phil@bubblehouse.org wrote:
I have to say, personally, there's pretty much no way I would ever name a project 'Twisty' anything ;-).
OK, OK, I get it. Everybody hates "Twisty" :). I should have been more careful to separate the specific suggestion of "Twisty" (which was just something that popped into my head) in my original message from the need for a word like this. I wasn't totally set on it.
Of course, I understand this wouldn't be a requirement, but if this is to be a useful marketing technique, whatever the prefix is needs to be obvious, but also inconsequential. That's the thing that's great about the Java convention, once you're familiar with it, you pretty much stop seeing the J. Same with the tired but successful iSomething.
You've convinced me. I still think "Twi" sounds okay (better than "Twisty") "Tx" (evocative of "TwistedmatriX", "Transmit", "Twisted multipleXed"?) "T"? I'd suggest "Tw" but I feel like it has to be pronounceable, and "Tw" forces the first letter of your project to be a vowel (whereas "Tx" could be pronounced "Tix").
I think a one- or two-letter prefix is pretty much the best option
OK. I guess your suggestion is a lower-case "t"? :)
I am curious, though, what would Divmod do? You guys have some of the most significant Twisted projects, and would probably be the best people to set the example, but you already have great names for your projects. tMantissa or twdAxiom just don't seem very stylin' ;-)...
Divmod is really enmeshed into the core Twisted community. If we had an implementation of protocol X for Twisted, we'd distribute it in one of our libraries or applications in order to prove it out, then contribute it to Twisted proper. We've done this a few times already. This sidesteps the naming issue, because then it's just "Twisted X" at the end of the day. In fact it usually isn't even that; we developed and contributed the IMAP protocol and that's part of twisted mail, we developed and contributed the JUICE protocol which became twisted.protocols.amp (although that will probably need to grow its own twisted.amp "dot product" at some point in the near future). I started trying to explain here why other people should do this if we don't, but it turned into a small novel about community dynamics. I think I'll save that for a future blog post. Suffice it to say that not everything that gets done in the world is as awesome as the stuff Divmod does, and naming conventions like this are for smaller, less heliocidal projects, which implement a protocol or two, or maybe bind to an existing non-Python library. And now, for something completely different:
On May 29, 2008, at 7:15 PM, glyph@divmod.com wrote:
I'd really like "twisted" (and our various "dot product" subprojects) to be a trademark that the software freedom conservancy can protect and defend.
A good example of what I wouldn't want to see is the whole Firefox/ Iceweasel debacle.
I was talking to James Knight about this yesterday (another core Twisted developer if the name does not immediately ring a bell) and he made a very good point about trademarks to me: "Tou especially don't want to discourage people from using the word "Twisted" to refer to Twisted." I think the Iceweasel thing is kind of silly, and isn't really necessary to protect the Firefox trademark, since Iceweasel *is* Firefox. If anything it weakens it a little bit, because now there are multiple ambiguous names which one can use to refer to Firefox, rather than one clear trademark. So I don't want anything like that to ever happen to Twisted. (When I say "weaken" here I'm not talking about the legal sense, but the branding / marketing sense. For all I know it might be bad for the legal sense too, but I'm not a lawyer.) On the other hand, Debian has recently had some rather high-profile examples of how they insert bugs into their packaging that are not present in the software itself[1], and Twisted has had issues with this too[2]. I can sympathize, a little bit, with the Mozilla folks' desire to use trademarks as a weapon to say "stop breaking our stuff!". I don't think that this is the right way to go about it, and clearly it doesn't work if they just do it anyway and give it a different name. While I think the Firefox patches which precipitated the Iceweasel debate are far, far from this point, there *is* some point where patching becomes so intense that distributors really shouldn't use the same name for their software. [1]: http://www.debian.org/security/2008/dsa-1571 in case you're living under some kind of magic rock that prevents all computer security related things from reaching you. [2]: I couldn't help but think of this bug - http://bugs.debian.org/cgi- bin/bugreport.cgi?bug=474630 - where some debian developers have specifically discussed re-introducing bug #2339 in a patch. I feel like there's a parallel thread that should be started here about how we communicate with packagers. For example, I am pretty sure that there are bugs in Twisted which would prevent *any* system packager (debian, redhat, gentoo) from getting a clean test run from an installed package, and yet we basically never hear about it. Why is that? What use is our test suite if the actual way twisted is installed always breaks it? How can we convince packagers to build and install stuff and tell us when it fails the tests? This has nothing whatsoever to do with trademarks, of course.