Remove user from queue if to slow

AlgoJerViA · Post by **AlgoJerViA** » 2005-11-19 05:54

Hi
I would like to see that it would be automaticly posibly to remove users from queue if theier download speed is not fast enough.
I have 24Mbit in and I often have to sitt and manually remove users from the queues in serveral diffrent files becouse theier upload speed is about 1Kb/s or even worse 100b/s... meaing the file i want to download wile take about a year insted of 10min.

//Andreas

joakim_tosteberg · Post by **joakim_tosteberg** » 2005-11-19 05:57

This have begun to get implemented in the current development version of DC++ and will be present in the next release.

bastya_elvtars · Post by **bastya_elvtars** » 2005-11-19 05:58

This (orsomething like this) will be implemented into DC++ 0.69.

AlgoJerViA · Post by **AlgoJerViA** » 2005-11-19 07:06

cool... is there a nighlty build? or is it posible to compile with Code::Block?

TheParanoidOne · Post by **TheParanoidOne** » 2005-11-19 07:12

There are no nightly builds. You can either compile it yourself from CVS code (though that isn't particularly stable right now) or wait for the next release.

Guitarm · Post by **Guitarm** » 2005-11-19 07:14

It's a CVS version which is available for browsing or downloading at http://sourceforge.net/cvs/?group_id=40287.

You need to have VS.Net 2003, STLPort and WTL to compile it.

Pothead · Post by **Pothead** » 2005-11-19 08:05

Isn't python a requirement as well now (for makedefs ?)

Guitarm · Post by **Guitarm** » 2005-11-19 08:10

Pothead wrote:Isn't python a requirement as well now (for makedefs ?)

I guess so, aswell as a CVS client for downloading the source (I use TortoiseCVS myself)

Wisp · Post by **Wisp** » 2005-12-07 15:12

Is multiple source downloading also implemented in the next version? Revconnect has that feature already and the source if freely available, so it shouldn't be hard to implement that until the next protocol is ready.

bastya_elvtars · Post by **bastya_elvtars** » 2005-12-07 15:23

Wisp wrote:Is multiple source downloading also implemented in the next version? Revconnect has that feature already and the source if freely available, so it shouldn't be hard to implement that until the next protocol is ready.

I doubt it will be implemented.

Todi · Post by **Todi** » 2005-12-07 15:29

Wisp wrote:Revconnect has that feature already and the source if freely available, so it shouldn't be hard to implement that until the next protocol is ready.

That's not quite the way DC++ does things. arnetheduck, the copyright owner of DC++, only implements code from mods if the authors write over the copyright to him, to avoid any nasty copyright-issues down the line (one holder = much easier to enforce copyright). So unless revconnect signs over the rights, which they're not likely to do i guess, you'll have to wait 'til arne makes his own version.

Wisp · Post by **Wisp** » 2005-12-07 15:41

Todi wrote:That's not quite the way DC++ does things. arnetheduck, the copyright owner of DC++, only implements code from mods if the authors write over the copyright to him, to avoid any nasty copyright-issues down the line (one holder = much easier to enforce copyright). So unless revconnect signs over the rights, which they're not likely to do i guess, you'll have to wait 'til arne makes his own version.

That's too bad. I thought that the GNU license allowed the usage of the code as long as the program where the code is used in, is also open-source.

Maybe some dc++ developer could take a look at the revconnect code and make something similar.

joakim_tosteberg · Post by **joakim_tosteberg** » 2005-12-07 15:46

Wisp wrote:
Todi wrote:That's not quite the way DC++ does things. arnetheduck, the copyright owner of DC++, only implements code from mods if the authors write over the copyright to him, to avoid any nasty copyright-issues down the line (one holder = much easier to enforce copyright). So unless revconnect signs over the rights, which they're not likely to do i guess, you'll have to wait 'til arne makes his own version.
That's too bad. I thought that the GNU license allowed the usage of the code as long as the program where the code is used in, is also open-source.

It does, but it is still the author of the code that has the copyright of it.

TheParanoidOne · Post by **TheParanoidOne** » 2005-12-07 18:00

Wisp wrote:I thought that the GNU license ...

License != copyright

liny · Post by **liny** » 2005-12-07 20:43

Todi wrote:arnetheduck, the copyright owner of DC++, only implements code from mods if the authors write over the copyright to him, to avoid any nasty copyright-issues down the line (one holder = much easier to enforce copyright). So unless revconnect signs over the rights, which they're not likely to do i guess

I would give copyright to arnetheduck, if he want.

bastya_elvtars · Post by **bastya_elvtars** » 2005-12-07 21:23

liny wrote:
Todi wrote:arnetheduck, the copyright owner of DC++, only implements code from mods if the authors write over the copyright to him, to avoid any nasty copyright-issues down the line (one holder = much easier to enforce copyright). So unless revconnect signs over the rights, which they're not likely to do i guess
I would give copyright to arnetheduck, if he want.

If only everyone thought the same way...

Carraya · Post by **Carraya** » 2005-12-08 03:21

I still don't think it's going to happen, even if liny would give the copyright away...

Post by **GargoyleMT** » 2005-12-08 12:22

liny wrote:I would give copyright to arnetheduck, if he want.

Interesting, I'll make sure he sees that.

iceman50 · Post by **iceman50** » 2005-12-09 20:41

hmm sounds interesting should be even more interesting to see what happens

Post by **GargoyleMT** » 2005-12-10 10:18

liny wrote:I would give copyright to arnetheduck, if he want.

I thought about this a bit more after I posted my response - you don't have copyright assignment as part of your terms for accepting patches, right? If not, copyright for the code in question is spread out over everyone who contributed to that code. In order to give the code to arne, you'll need permission from everyone who patched the multi-source code. I think BigMuscle did, and he'd probably be receptive to it, but you still have to clear it by him and everyone else.

Big Muscle · Post by **Big Muscle** » 2005-12-16 15:41

I didn't patched segmented downloading in RevConnect. I've only improved in StrongDC++ to use ADCGET command, download by chunks and not need reconnect when chunk is finished.

If arne wants it, he can take it.

phaedrus · Post by **phaedrus** » 2005-12-20 14:46

the way i see it, if arne does this, he will single-handedly turn the entire dc comunity into a segment downloading comunity. his client right now is the dam holding back the entire ocean. it would bring a paradigm change across the entire comunity, the conversion is no small leap to take. i love the convenience of segment downloading but i dont think it would be such a good idea having it adopted in such a massive scale. if it does get implemented, it should not be built upon a single segment framework.

Todi · Post by **Todi** » 2005-12-20 16:29

Perhaps, if it's implemented, it should be ADC only.. in order to encourage a protocol shift.

Post by **GargoyleMT** » 2005-12-20 16:41

Todi wrote:Perhaps, if it's implemented, it should be ADC only.. in order to encourage a protocol shift.

The parts that make it (safely) possible already are borrowed from ADC - hash exchange is done only with $ADCGET's name spaces. Blocks of the file can be requested with $GetZBlock/$UGetBlock/$UGetZBlock, which is nominally NMDC, even just as an extension.

Post by **GargoyleMT** » 2005-12-20 17:53

phaedrus wrote:the way i see it, if arne does this, he will single-handedly turn the entire dc comunity into a segment downloading comunity. his client right now is the dam holding back the entire ocean. it would bring a paradigm change across the entire comunity, the conversion is no small leap to take. i love the convenience of segment downloading but i dont think it would be such a good idea having it adopted in such a massive scale. if it does get implemented, it should not be built upon a single segment framework.

So it's fine for you, but not okay if everyone does as you do?

Regardless, we usually talk about upload queues and credit systems when we talk about segmented downloads, at least historically. I don't buy the "the sky is falling" argument; I'm reserving my judgement.

PseudonympH · Post by **PseudonympH** » 2005-12-20 18:54

At least for an initial implementation, the multisource should try to not be too greedy -- it should limit itself to a handful of sources, especially if multiple files are being downloaded. Once we see the effects of it being widespread, then we can move from there.

cologic · Post by **cologic** » 2005-12-20 18:55

That might prove useful anyway, under the XP SP2 TCP/IP stack.

RHIncredible · Post by **RHIncredible** » 2006-01-13 22:26

All this talk about segmented downloads and copyrights is fascinating but honestly I would REALLY love to see this auto-kick for slow downloads implemented. I can not express just how frustrating it is to sit and baby sit my queue because there are 500 users with the file I want and 450 of them can't even sustain 1KB/s and then once one of these gracious users with 1KB/s is found with a slot for that file I miss out on all the decent connections that I could be downloading from if it wasnt already being downloaded from this crappy connection. I would MUCH rather just wait till a reasonable connection is available rather than download files in this dribble fashion.

Todi · Post by **Todi** » 2006-01-13 22:58

RHIncredible wrote:All this talk about segmented downloads and copyrights is fascinating but honestly.. blabla

Bother to read the changelog for DC++ 0.68 before you complain?

RHIncredible · Post by **RHIncredible** » 2006-01-14 08:22

Todi wrote:
RHIncredible wrote:All this talk about segmented downloads and copyrights is fascinating but honestly.. blabla
Bother to read the changelog for DC++ 0.68 before you complain?

Actually I wasnt complaining I was expressing my enthusiasm for an invaluable feature, also I was using .674 and since the posts above said that the feature was going to be in .69 I didnt bother cheching .68, now I have.

SubversiveAgent · Post by **SubversiveAgent** » 2006-03-27 07:39

Seems I wasn't the only complaining. Today I complain and find out the feature already got implemented in the last build. Now THAT is good service!