[deimar]

I'm curious about how the features purposed in this forum will be viewed by the development team and how much they'll consider them. Anyone can answer me?

Anyway, I have some purposes about the comments, since they're somehow vital (for reasons I don't even need to explain) for emule and all its users.

1) A quick connection from a just-downloaded file in the "Transfers" list to the corresponding name in the "Shared Files" list, so that a person can quickly find the file he wants to comment. This would be important because people often download more than a file for the same thing, just to find the right one, given that nowadays there are 4-5 fakes for every good file. So you don't go crazy every time you have to comment many files with almost the same name.

2) A way to keep on the client computer (or on a server or wherever you want) the comments of cancelled downloads. This because you have to keep the file if you want to keep the comment and the file you downloaded takes up space on the Hard Disk. (Just yesterday I discovered I had 74 GB of unfinished downloads that I had stopped and kept untill then just to grant my comments to other users)

3) A small popup which asks to comment the file when the download is finished would make even the laziest or te most incapable people partecipate to the community with comments.

Thank you

[fox88]

1) I disagree with "4-5 fakes for every good file". Besides, comments were not designed to be 'fakes detector'. Comments must be treated with caution; and some of those are just spam.

2) You can be reasonably sure it would not happen in ed2k; and I do not think it would be implemented in KAD. There is only a limited number of files which user can share; and every 'shared comment' would increase traffic and decrease the number of shared files.
What's more, it would be useless to keep sharing comments for files which are not shared anymore. Then you'll have to think about the time limit for sharing comments - or even about some kind of manual 'comments management system'.

3) eMule does work without any user attention most of the time. Your popup would be hanging around for hours. But even if you see a popup right after completion of download - what kind of comment you can give without even looking at the file? It would take 1-2-3 hours to watch a film, for example. Hence popups would be simply an annoyance.

[deimar]

Thank you for your answer fox88, it was very clear and straight. Nevertheless you have to admit that there are many fakes in both kad and ed2k, and comments are a good way to slow down their spread. Besides, if one of the choices that you can choose when you comment a file is "Fake/...", it means that at least they've thought about this possibility...
And last, I fail to see how the traffic generated from a string of comment can be the same as the one generated by a shared file. This rule of the one file = one comment seems to me a little silly, but I'm not a professional of nets and I will not reply on this.
But you'll agree that comments are a very powerful tool for users. I hope that in future we'll see the comments improved in Emule, and all these tecnical problems solved towards this objective.

[deimar]

fox88, on 21 November 2011 - 07:56 AM, said:

I disagree with "4-5 fakes for every good file". Besides, comments were not designed to be 'fakes detector'.

I'd like to specify that the 74 GB of unfinished downloads were all fakes that I had stopped after having seen the comments.
So, in my opinion
- the problem of the space is real in the case of the fakes
and
- comments do help to avoid the spread of these fakes, even though they were not meant just to do that.

[Nissenice]

Sorry, but it's not as easy as you seem to think it is.

Afaik when you make a comment on a file there is actually no check that you actually share it. Nor is it any check that you or your comment is to be trusted.
This means that whenever you make a comment about a file you consider as a fake file there might be hundreds of comments (probably faked) announcing the same file as good.
Or in the case you have made a comment about a file as good there might be hundreds of others comments marking the same file (falsely, of course) as for instance 'child pr0n'.

So in short, there is no check that published files are what they seem to be and there is no check that comments actually describes the true nature of the file. So in the end if many peers started to comment files we would not only have faked files, but also faked comments.

So, in that case, what have we actually gained?

[deimar]

Nissenice, on 24 November 2011 - 05:32 PM, said:

So in short, there is no check that published files are what they seem to be and there is no check that comments actually describes the true nature of the file. So in the end if many peers started to comment files we would not only have faked files, but also faked comments.

So, in that case, what have we actually gained?

So, if comments can't be checked nor trusted, what were they made for? In my personal experience, I've always found just useful comments, and very few were useless. IMHO comments have teir own value and if a way to control them and "guide" them was ever found, we would have a better p2p then.

[Nissenice]

deimar, on 27 November 2011 - 05:15 PM, said:

So, if comments can't be checked nor trusted, what were they made for? In my personal experience, I've always found just useful comments, and very few were useless. IMHO comments have teir own value and if a way to control them and "guide" them was ever found, we would have a better p2p then.

Well, I don't think I was around here by that time. But I can imagine there was a feature request about it and at the time it was found useful. I suspect there were hardly any fake files or index pollution at the time this feature was implemented. Nor any comment-spamming clients from e.g. Israel, which I assume appeared later, which exploited this feature to advertise their own mods. The latter explains the comment spam filter in Options/Messages and comments, I guess.

Quote

In my personal experience, I've always found just useful comments, and very few were useless.

That's because not enough people use the feature. If many people would start to use comments to warn about fake files distributed by AP2P-orgs they would in turn reply with polluting the comments with faked comments. At the moment they don't need to spend any resources on this because there wouldn't be any gain.

Quote

IMHO comments have teir own value and if a way to control them and "guide" them was ever found, we would have a better p2p then.

Control either means use of central server(s) and that's out of the question because their 'single point of failure'. We all know what have happened to the ed2k-servers. No?
Or it means use of a reputation system. If it exists one worth naming, I fear that's out of the question as well because of e.g. the extra overhead.
Another side I think is worth mentioning is that the more secure and verified information would get, the more the same information would be trusted as forensic evidence.

Personally I think the most reasonable way to go is to make use of ad hoc, heuristics, ipfilters to reduce the pollution in the network.

This post has been edited by Nissenice: 27 November 2011 - 07:19 PM

[deimar]

Ok thank you for your explainations, I have no other observation.
We just have to wait to live in a better world.

[Omnithec]

I'm sorry to bring this up again but, I noticed the only reason for comments discussed here was how they could be used to distinguish between fake and real files.

Is that really the only use for them?

There are other very good reasons to use comments.

Have you noticed how big the name of the files become to let people know important information about them?

I can give you some ideas where this might happen.

What if someone has the care to let people know what the file is with detail (and this is a good thing) and organize them with similar names when they are related (also a good thing)?

The path size limit on a modern windows system is 255 characters. Some times it is impossible to put all the information that we would like to place on the file's name. This isn't just because there can be a lot of information. I can also be because there is enough information to make the name just big enough to be too big when added to the name of the directories where it is. It's 255 characters for the entire path, not just the file name.

This may not be the most important part. Although this can happen more easily than you might think if people start taking sharing a bit more seriously and want to make life easier for other peers.

What is important is that there are also other good reasons to want space to describe the content. This could even be real metadata that the peer can't access because he would need a specific part of the file to reach it.

If the comment was in the form of a null terminated string and we wanted a nice round number (as far as binary goes) that could be done by using exactly 1024 bytes (1KB) but it could be much smaller. Even 255 characters would be a great help. But let's assume we decide to go with 1024B. Since we would have to have one byte for the null character, the comment could reach a size of 1023 characters. Remember that the comments would only be transmitted once per file per search.

If someone had a list of 1024 files the comments would occupy 1MB. This might seem much, but remember this is just text! It is super easy to compress it to a ridiculously small size! Locally the comments might occupy a couple of Mbs but when someone wanted to see someone else's list of files, the comments would be transmitted in a compressed form that wouldn't make much of a difference.

But even this is only if we used 1023 letters. If we used 255 letters it would still be useful and, with compression, would make almost no difference.

If there are more than one comment for the exact same file in a common search, instead of a peer's list of files, the peer could see those that appeared in his search and select one. It isn't even necessary to search for every comment available. Just get what's at hand, what appears in a normal search, to avoid getting a bigger increase in overhead.

IMHO it could be very useful to have proper comments linked to the files with part of the content description, capable of being transferred from peer to peer and even changed by each one if that becomes seen as practical (sometimes other peers might know something new about the content and want to add it). If 1024B is too much, why not 256B? It would still be useful.

There should be an easy way to consult them. I would like to have a small popup appear with the comment when the mouse was left over the file name or I selected an option in the context menu. There should also be a way to turn it off if someone finds it too annoying or simply doesn't care about that.

Wouldn't it be nice?

[Link64]

Omnithec, on 04 February 2012 - 06:01 PM, said:

The path size limit on a modern windows system is 255 characters.

No, it's not.

From Wikipedia/NTFS:

Quote

File names are limited to 255 UTF-16 code points. (...) The NT kernel limits full paths to 32,767 UTF-16 code points.

[Omnithec]

My mistake. The reason I thought it wasn't is because I have never been able to have file paths bigger than that.

I thought it strange that it couldn't be done, so I am not that surprised I was wrong, but the fact is windows 7 is always complaining the path is bigger than 255 characters when I try to create one.

If you know what's happening to keep me from having file names bigger than 255 characters, please tell me.

Anyway, I still think it would be useful to have a comment. Having everything crammed into the file name can be uncomfortable. Even if the name can be bigger, it would still be more practical to have comments to help understand what it is we are actually downloading. From the comment, people could decide by themselves what information to keep.

You may say that the comments would inevitably be falsified but that already happens with file names. There would be no increase in fake files, there would only be an increase in the quality of the description of real files.

I hope you agree with the rest of my post and realise that. I would like to know what you think about it, not just this correction (as important as it may be).

[Link64]

Omnithec, on 05 February 2012 - 12:55 PM, said:

If you know what's happening to keep me from having file names bigger than 255 characters, please tell me.

The red part...

Link64, on 04 February 2012 - 06:31 PM, said:

Quote

File names are limited to 255 UTF-16 code points. (...) The NT kernel limits full paths to 32,767 UTF-16 code points.

Anyway, we already have comments, basically nobody uses that so far and actually I doubt that even if we had an improved comment system, anything would change. If someone likes to share detailed information about a file, which does not fit into the current comments, sharing an .nfo file with the same name might be better anyway, as you don't have all that information in one row, so it's a lot better readable.

[Omnithec]

Hmm... You gave me food for thought, but...

Don't the comments we add stay with us and are not sent along with the files?

If those comments already move along with the file, then I don't understand why this thread started in the first place.

If they are not, I think you might be right about people not using them... but perhaps not as much as you might think.

One thing for which I search a lot is books. A lot of other stuff too but one thing I have plenty of are books. I don't know how it goes for films and others but, in many cases, the book names have some sort of extra information. I have over 45 000 digital books to prove it. Many are different versions of the same book, and I admit that obviously I will never organize most of them (which doesn't mean that most won't be used, though it is true that many are there "just in case"), but their names vary and their proportions are balanced.

That information appears mostly on particular kinds of books but it does appear. That means that someone went to the trouble of adding metadata to the names of around 1/3 of the books. Possibly more, but I rather be mistaken by less than by more. That means 15 000 books. It is easy for me to be wrong about the right proportion, but I have my experience to help and I have just used a third party program to search for *.* inside the folders where the books are and a quick look at the list shows it seems to be roughly right.

Those who share certain types of files, with certain types of content, seem to like to add metadata. I think you are right that many (or most) wouldn't care, but this might change over time. You see, this doesn't happen only to books and the comments wouldn't be wasted. Even if only a few added comments, you must remember that they would not go away, at least not the good ones. This means that over time more and more files would have comments.

Who knows, maybe some users would get a kick from being the first to comment a less known file with freedom to write what they want. Remember, bad comments would be as frequent as bad names are now and would probably have a short life cycle. Users would be able to choose between different comments and get the one that seemed to be the best, and I don't think those would be the ones of bad quality.

There are things that are better left on the name because they are sufficiently important to have to be guaranteed. There are also things that only matter when choosing the files you download and/or while the files are being shared. Even that information which really has to be left in the name, to make sure it isn't lost, could be better explained in a comment at the same time.

And about their use. I'm sure everyone would start to know and use this option in a passive way, even if only a few would do it actively. Forums, support/help, FAQ, word of mouth, even notes like “see comments” in the very name of the files would lead to the dissemination of the use of comments.

Moving on.

The nfo option is something I thought about with txt when I used another network a long, long time ago. The problem is, those files wouldn't be selected for download. You must admit those files would feel a lot less practical than comments. In that case, I'm that no one would want to use the files. Also, if more files started to be shared because of that, there would be more processing and bandwidth allocation (while comments would only be a matter of copy-paste). It might also diminish the amount of true files. This would happen for at least two possible reasons: because the users would have to sacrifice space in their file list, to accommodate those files without creating the increase in resource use I mentioned before; those files would be published and would diminish the amount of useful files visible to the other users.

[Link64]

Omnithec, on 05 February 2012 - 07:02 PM, said:

Don't the comments we add stay with us and are not sent along with the files?

If those comments already move along with the file, then I don't understand why this thread started in the first place.

They are send to every client which requests a shared file with a comment. The point (the 2nd actually) of this thread was to share the comments after the file is not anymore in the share for whatever reason.

Omnithec, on 05 February 2012 - 07:02 PM, said:

One thing for which I search a lot is books. A lot of other stuff too but one thing I have plenty of are books. I don't know how it goes for films and others but, in many cases, the book names have some sort of extra information.

Films and other stuff uses the same naming schemes, i.e. something like "Title.720p.AC3.BDRip-ReleaseGroup.mkv", eventually with more information about language or whatever. Sure, not everyone follows that and that's IMO the actuall problem. If anybody wants to do something good, better name that file properly (and hope others will choose to use that name as well) instead of adding a comment which will be gone once we unshare the file.

Omnithec, on 05 February 2012 - 07:02 PM, said:

I'm sure everyone would start to know and use this option in a passive way, even if only a few would do it actively.

They can use proper file names in a passive way as well and eMule "supports" that already.

Omnithec, on 05 February 2012 - 07:02 PM, said:

The nfo option is something I thought about with txt when I used another network a long, long time ago. The problem is, those files wouldn't be selected for download. You must admit those files would feel a lot less practical than comments. In that case, I'm that no one would want to use the files. Also, if more files started to be shared because of that, there would be more processing and bandwidth allocation (while comments would only be a matter of copy-paste). It might also diminish the amount of true files. This would happen for at least two possible reasons: because the users would have to sacrifice space in their file list, to accommodate those files without creating the increase in resource use I mentioned before; those files would be published and would diminish the amount of useful files visible to the other users.

Sure, additional files are not optimal solution for a network like that, but publishing persistent comments as requested in this thread would generate about the same load while being less useful, unless you basically allow to publish the entire nfo file as a comment, i.e. we would need the possibility to have more than one row. If we are limited to one row, IMO better stick to proper file names, 255 should be enough for all relevant info.

[Omnithec]

If comments are already transmitted with the files, there is no problem.

If that is the case, there is no point in proceeding with this debate, as far as this particular idea is concerned.

I had a few arguments to share, and I think my position is valid, but ok, there's no point now.

I'm taking your word that it works that way. I'm going to put “[see comments]” wherever I cannot adequately describe the file and need to place extra information.

So, I hope you're right.

-- Edit --

Oh, and about the point you mentioned about keeping the comments around, the one I misunderstood for this one. I think it might be useful, at least in some ways, but is far from feasible and would not be properly used even if it was possible.

This post has been edited by Omnithec: 08 February 2012 - 03:21 PM

[fox88]

Link64, on 06 February 2012 - 02:46 AM, said:

If anybody wants to do something good, better name that file properly (and hope others will choose to use that name as well) instead of adding a comment which will be gone once we unshare the file.

However, it does not address the main issue (#2 from the first post). Naming fake as "fake.avi" (or "virus") is even poorer than comments: still works only while the file is being shared, but there would be no warning icon in files list.

Link64, on 06 February 2012 - 02:46 AM, said:

publishing persistent comments as requested in this thread would generate about the same load while being less useful

To be precise, currently comments do not require pulishing yet another file; and we know that eMule can effectively publish only a limited number of files.

[Link64]

fox88, on 09 February 2012 - 08:09 AM, said:

Link64, on 06 February 2012 - 02:46 AM, said:

If anybody wants to do something good, better name that file properly (and hope others will choose to use that name as well) instead of adding a comment which will be gone once we unshare the file.

However, it does not address the main issue (#2 from the first post). Naming fake as "fake.avi" (or "virus") is even poorer than comments: still works only while the file is being shared, but there would be no warning icon in files list.

That was intended for good files, Omnithec was talking about 255 chracters in file name not being enough to describe the content of the file. In real world fakes and viruses are deleted as soon as the user knows about them, basically nobody will keep an infected file on his computer just to inform others about that.

fox88, on 09 February 2012 - 08:09 AM, said:

To be precise, currently comments do not require pulishing yet another file; and we know that eMule can effectively publish only a limited number of files.

The current comments don't, but #2 from the first post requests persitent comments, which would have to be published some other way once the corresponding file is deleted localy.

[Omnithec]

I don't think it is a good idea to have comments wander around the servers. Actually, I think it's a terrible idea.

All I wanted was proper comments attached to files, not uploaded into servers where they would create havoc.

Besides the point that one of the objectives of eD2K is to avoid keeping information on the servers, your idea isn't feasible.

Think about it: whose comments would be saved? The ones made by the first user that shared the file? Every comment from everyone?

If the comments are made by only one person or a group of persons, they are useless because there will not be a way to impose a good overall bias towards a majority of truthful comments.

If everyone can add comments, have you thought about what that would mean in terms of server operation? What it would mean to the working of the network?

Even if the network could miraculously work with all of that, how could someone understand what's what with so many comments to read?

[fox88]

Link64, on 09 February 2012 - 12:20 PM, said:

In real world fakes and viruses are deleted as soon as the user knows about them, basically nobody will keep an infected file on his computer just to inform others about that.

I'm afraid you did not read the whole point #2 in the very first message of this topic; something about 74G of stopped files.

Link64, on 09 February 2012 - 12:20 PM, said:

The current comments don't, but #2 from the first post requests persitent comments, which would have to be published some other way once the corresponding file is deleted localy.

Not necessarily.
Though I still do not believe in practically usable implementation.

[Link64]

Omnithec, on 09 February 2012 - 08:01 PM, said:

I don't think it is a good idea to have comments wander around the servers. Actually, I think it's a terrible idea.

All I wanted was proper comments attached to files, not uploaded into servers where they would create havoc.

Besides the point that one of the objectives of eD2K is to avoid keeping information on the servers, your idea isn't feasible.

Think about it: whose comments would be saved? The ones made by the first user that shared the file? Every comment from everyone?

If the comments are made by only one person or a group of persons, they are useless because there will not be a way to impose a good overall bias towards a majority of truthful comments.

If everyone can add comments, have you thought about what that would mean in terms of server operation? What it would mean to the working of the network?

Even if the network could miraculously work with all of that, how could someone understand what's what with so many comments to read?

The comments would not be saved on servers, servers can't do that and their development was stopped few years ago. So it could be only implemented in Kad. Also the comments would not be saved anywhere, but published in Kad by the user who wrote them, so they would be available as long as the user is online or untill he deletes them, i.e. you could keep your comment available to the network without keeping the file. That was more or less the reqest from post #1.

fox88, on 10 February 2012 - 07:16 AM, said:

Link64, on 09 February 2012 - 12:20 PM, said:

In real world fakes and viruses are deleted as soon as the user knows about them, basically nobody will keep an infected file on his computer just to inform others about that.

I'm afraid you did not read the whole point #2 in the very first message of this topic; something about 74G of stopped files.

Yeah... I did read that, but how many users out of all do it that way? I mean, most of the time I could even not afford 74GB of my hard drive just to keep my comments available and what's worse, use my bandwidth to upload it to others.