Shareaza

I've been trying to download a file for months now and it's gotten as high as 99.5% complete but then for no discernible reason it drops back down to around 98.25%. Last night it got to 99.33%, then it stopped downloading before I shut down and went to bed. Today I started up Shareaza and it was still at 99.33% but since then it's dropped down to 98.69%. There are 102 sources. Any suggestions on how to finally complete this download? Or any idea as to why it won't finish? Thanks.

There are files where no complete copy exists on the network. If every one of the hundred sources of your file misses the same piece of information, it is possible that your download never completes. As for the dropping in percentage, that's normal: if Shareaza detects that it has downloaded some corrupted data (e.g. some parts of the file are not as they should be), it redownloads them, to avoid corrupt files that might be damaged / unplayable etc.

I don't think you got the point: let's say you are sharing a movie with some 100 MB of data. Now while you are sharing it, lots of people decide that they want to have the file, too and they start requesting parts of the file from you. Now you've uploaded 99,5% of the file to these users (which means that there are 0,5% of the file that none on the network besides you got on his PC) and decide to unshare the file, so these 0,5% (or whatever percentage hasn't been uploaded) will never be downloaded by anyone.
Now eDonkey lacks a way of telling whether there is a complete copy of the file. This means that others could still find that file even though it is not available as a complete copy and start downloading it. And then you get at a point where there are suddenly 100 people trying to download that file - however, most they can ever get is 99,5% of it, so that download will never finish.

As for the repeated redownloading of some chunk(s), there is a simple explanation for that, too: Shareaza uses very sophisticated ways of checking that the data it downloads from others is valid. However, not every client has these methods in the same way we do. So just imagine there is one client among those you download from that has a piece of data you don't have, however that piece of data is corrupt, but hasn't been validated correctly so the client assumes it to be OK. Now when this client uploads a part of the corrupted data (in fact, it only needs to upload 1 single corrupted bit) to you, Shareaza needs to redownload the entire chunk, and there is no way to prevent this, as Shareaza cannot tell whether a specific bit is OK or not, it can only validate the complete chunk. (If we wanted to check every single bit of a download while transferring it, we would need to download the the file two times, and this would multiply the amount of time it takes to download a file by 4.)

Just an additional point: the hashing is done on completion of file for the file hashes and on the completion of a chunk for the verification hashes and only verified chunks are shared. But there is no mechanism to detect file corruption after hashing... So corrupt chunks do sometimes get shared. And of course, corruption between downloading and completion will look like the source sent a bad chunk even if the corruption happens on the downloading machine or in transit.

offtopic:
I think this thing about telling people they are misunderstanding or mistaken about something being rude or insulting must be some sort of a cultural difference. So since I don't actually want to be rude, what is the polite way to tell this? Do you simply omit using the "bad" words and tell what was wrong and what is correct? (Even I could manage that.) Or is it about using "that is wrong" instead of "you are wrong"? Advice would be greatly welcome. And does anyone know why and when have these expressions become rude?

An explanation: Where I live humans are considered innately fallible creatures. "Errare humanum est" and Christian theology and all that... So saying someone is wrong, mistaken, or misunderstanding is roughly as insulting as saying they are breathing. Part of being human.

OK I think I understand what everyone is suggesting however when I expand the sources there are at least 3 sources that Shareaza indicates have the full files to download. They all began downloading at decent speeds got up to around 99.6% and then all the sources dropped off and the progress drops back down to the mid-98%...

The idea that a majority of these 102 "sources" don't have the chunks I'm missing isn't too far fetched. I've been sharing the partial files for months now and I've noticed a few of the sources have pretty much (possible exactly) the same missing chunks on their progress bars; so they could have downloaded all of their copy of the file from me and are in the same hole I find myself.

Let me see if I understand a point that was made early... Are you (old_death) suggesting that the users with complete files may have some kind of "corrupt" data in my missing "chunks" that they were able to download but I and others like me are able to download them but then are detected or interpreted as corrupt so I dump the data? If so, trying to complete this download is futile since, as was suggested, if it were a different file it would be hashed separately, so all the complete files I'm trying to download will inevitably fail.

thanks again

Speak for yourself. My brain is working just fine, thank you.

Never claimed otherwise. Just saying that the "working fine" of the brain or any other neural network is not quite the same thing as that of conventional computers and might not mean what people assume it to mean. Incidentally, the question about the limitations of human knowledge or understanding apart from being one of the main topics of philosophers for millennia actually has become quite important in recent years. Issues such as the climate change and GM crops have brought science into the public spotlight and much (most really) of what the scientists do is about catching smart, intelligent, and dedicated people being wrong and then correcting the errors. Anyone assuming being wrong implies either incompetence or dishonesty will end up having a very dim view of scientists and what they say when exposed to that. And obviously assuming that we either do or do not know something will make the statements the scientists actually make with their mathematically accurate but rather arcane confidence levels seem unreliable and "iffy" when in fact the particular data may be more reliable than the "everyday knowledge" that is accepted without question.

Just to clarify things -- I don't think guessing wrong indicates incompetence. It's not knowing it's a guess, or being wrong when it's not supposed to be a guess, that indicates incompetence or, sometimes, that one has been lied to (including possibly by omission and without strong intent, if someone told their guess to you as if it was a known fact rather than a guess, and then you repeated it).

It is logically impossible to know things you do not know. That means you have no way to know whether something you do not know would be important. Consequently, your information may always be insufficient without you knowing it regardless of the amount of information you have. There are methods used by scientists to calculate confidence levels that basically tell the probability of there being something they don't know that would change the result, but they require significant effort, discipline, and expertise to use and never will get you a certainty just a probability.

I am not one of those doom and gloom the end is near people, but does the world we live in really look like people have some practical way of knowing in advance when they are wrong? I suppose you could assume original sin and assume people are wrong because somebody lied far enough back in time that it now affects everything. (I think many people do believe this, actually.) But that would simply change the reason people are wrong without knowing it, not the fact they are.

Actually, the point was that while it is possible to know you are "guessing" instead of "knowing", it is logically impossible to know you actually "know" something instead of just "guessing". In fact, what we usually mean when we say we "know" something is that we have a high confidence in it being true, while "guessing" means we have a low confidence in it being true. There is no qualitative difference between the two kinds of knowledge, things we have high confidence in may be false and things we have low confidence in may be true. There is also no magical oracle telling us how reliable what we think is. The confidence we have is simply an estimate the brain automatically makes based on the information we have. Basically instead of telling how likely something is to be true, our "knowing" is based on how consistent something is with the information we have. This estimate itself is usually highly unreliable, but we are not generally aware of that as it is something the brain generates automatically.

uh guys? can we try and stay on topic please?

How about including a memory tester in Shareaza package and ask to run it in first run wizard?

Not very useful. As tharrison1 said those problems are pretty rare. And memory testers don't usually work reliably under Windows and instead have to be run from boot. (Recent versions of Windows actually have one included that is run automatically if Windows suspects memory instability.) And anyway even if you find a good memory tester that runs without requiring a reboot, testing the memory takes time. You need to test every memory address with several test patterns multiple times each. (Since some problems only show with some patterns and if the RAM is warm enough.) So it would be lots of hassle for little gain running one for every install.

Oddly enough, the best free "memory tester" that runs as a normal Windows application that I found was... Shareaza. Both P2P downloading and hash verification turned out to be very sensitive to memory problems. So it actually would be pretty efficient to just keep track of verification errors and if you have within a certain time sufficient number of errors from different peers and files, show the user a pop-up saying that there is probably a problem with his system.