[NNTP] Compressed LIST (and other commands) answers

[Sabahattin Gucukoglu]

On 2 Dec 2009, at 10:16, Ade Lovett wrote:
> On Dec 02, 2009, at 02:07 , Urs Janßen wrote:
>> JFTR giganews has implemented something like this:
> 
> I am fully aware of what Giganews has implemented.
> 
> It's a crock.  Plain and simple.  If they had _any_ respect for the community of which they are a part, they'd have offered up some form of communication on this list.
> 
> The fact remains that they've chosen to implement some random ad-hoc solution, no doubt driven by just-out-of-school business development folks, surrounded by a "the family can do no wrong" attitude, to implement something that, quite frankly, is complete garbage.
> 
Sounds like a whole lot of very sour grapes you have there.  I don't see any logical connection between quality of implementation and respect for standardisation; the two tend to be beneficial to each other but are basically exclusive.  I would much rather you told us flat out how big bad Giganews made your life miserable by being so successful and yet, like so many large companies, such obvious sharks.

To be clear: I am a customer of theirs because I like their service (read: lots of browsable text news with awesome retention on the lowest plan).  And yes, compression helps; large listings are constructed and compressed once and then consumed many many many times.

I am well aware of the, erm, roughness of the implementation, though I hasten not to call it "A crock" and "Garbage".  It serves their needs.  I would add compression on text-only groups, myself.  My request for documentation, in order to in some way appease the many non-Windows (their proxy is Windows only) users, never made it to the blog comments on their oh-so-fabulous proxy accelerator promotion.  But Newsbin, a popular binary grabber for Windows, managed just fine without it.

> Ask yourselves these two questions:
> 
> 1.  If I'm attempting to download binaries faster, what does <compression-technique> give me?

Nothing.

> 2.  If I'm spending a lot of time in high-traffic text newsgroups, what does <compression-technique> give me?
> 
Everything.

> Pay particular attention to the fact that consumers of (1) and (2) above are most likely already on high speed internet connections (heck, I pay $19.95/mo for 10/3) and so are _not_ the target audience.
> 
Whether you benefit depends on whether the bottleneck is CPU or network.  On my Pentium II, compression is not welcome; on my dual-core MacBook, it sure wouldn't hurt.

(I was lying when I said "Nothing" or "Everything".)

> Again, I ask folks.  Is this really a "problem" that needs to be "solved" ?

If we can get past all the loathing, I'd say it wasn't a *serious* consideration, but for fear of multiple implementations doing the same thing in many stupid ways, I think we should try our hardest to best the lot of them.  Excellence is, after all, what the IETF is good for.  And when we have those specifications, we can legitimately point to the NSPs who ought to know better, and to the implementers who will appreciate the one right way, and bring them aboard for features people will actually like.  Right now, it's a bit hard to deny that compression is a popular idea that might just have merit.

Cheers,
Sabahattin