[2011-03-12 12:58] <cologic> The other issue I've heard is multisource downloading, which I intentionally didn't point out that could be disabled in the blog post (don't want to broadcast encouragement) but have told individuals freely
[2011-03-12 12:59] <eMTee> yeah, it can be disabled so its not a valid point against the upgrade
[2011-03-12 12:59] <cologic> Right...
[2011-03-12 13:00] <eMTee> however it could be still improved, but arne doesn't accepted my reasons/theoretic proposals
[2011-03-12 13:00] <cologic> such as?
[2011-03-12 13:01] <eMTee> as I see in the changelog, Eiskalt just added a setting to fix the segment size, which is not the good solution but at least a remedy for those who are hit by the disadvantage
[2011-03-12 13:02] <cologic> Oh... Indeed. A minimum segment size at least is pretty defensible
[2011-03-12 13:02] <cologic> But really, it should just have a better default
[2011-03-12 13:02] <eMTee> cologic, its simply works bad in very low quality connections
[2011-03-12 13:02] <cologic> define 'very low quality'
[2011-03-12 13:02] <eMTee> ok then you see my claim about it
[2011-03-12 13:04] <eMTee> when eg. a 3rd world isp has slow/bad/large ping value connections to outside the country
[2011-03-12 13:04] <cologic> none of the popular DC countries really qualify as 3rd world... they often have pretty good ISPs
[2011-03-12 13:05] <eMTee> the dynamic segment size goes to the minimum (64 or 128 KiB) in this case but in these connections the 'Connecting...' time is really large
[2011-03-12 13:05] <cologic> Oh, latency not bandwidth
[2011-03-12 13:06] <eMTee> yeah, that's the good word
[2011-03-12 13:06] <eMTee> smaller segments -> more connecting phase
[2011-03-12 13:06] <cologic> Segments that small are pointless anyway, mostly
[2011-03-12 13:06] <cologic> The minimum should be at least, say, 512kB
[2011-03-12 13:07] <eMTee> download time ends up 10x slower than it would possible
[2011-03-12 13:07] <eMTee> I seen this many times with my own eyes
[2011-03-12 13:07] <eMTee> I recommended 1MiB as a minimum
[2011-03-12 13:07] <cologic> also reasonable
[2011-03-12 13:07] <cologic> asked arne, or is that what he's responded to?
[2011-03-12 13:08] <eMTee> I don't remember exactly. must find it in the logs
[2011-03-12 13:08] <cologic> I can understand not wanting to add a setting, but adjusting a broken minimum should be less controversal
[2011-03-12 13:08] <cologic> *controversial
[2011-03-12 13:08] <eMTee> I didn't recommended a setting
[2011-03-12 13:09] <eMTee> min segment size 1MiB or less if the file is smaller than 1MiB
[2011-03-12 13:09] <eMTee> probable 2 lines of change in the code
[2011-03-12 13:09] <cologic> sure, presumably the issue is that someone (arne?) might object, not that it's hard
[2011-03-12 13:09] <eMTee> I don't see any disadvantage for good connections
[2011-03-12 13:09] <cologic> nor I
[2011-03-12 13:11] <cologic> maybe canvas the hub for opinions on this?
[2011-03-12 13:11] <cologic> I mean, popularity isn't a sound argument but it can be an effective one :p
[2011-03-12 13:18] <iceman50> i think it's good
[2011-03-12 13:19] <iceman50> no point in having small segments anyways
[2011-03-12 13:19] <iceman50> at least not *that* small