IP Filter. Yes or No ?

[akadajet]

Don't waste your time developing this. It just creates a false sense of security and needless bloat.

[nerdnamedmark]

False sense of security, false positives. It's just a bad idea.

[silverfire]

It's my opinion that if you have to worry about things like this, stop downloading illegal things from peer-to-peer applications.There are better programs for the job

[ScubaSteve]

/me applauds the last 6 comments

[Vicsun]

I registered just so that I could vote.

To paraphrase a popular sentiment: blocking peers flooding the network with bad data is a-okay, but implementing a feature which is possibly useless to everyone and certainly useless to those who use bittorrent legitimately is a horrible idea. Please don't do it

[Bad Seed]

Blocking IPs of people who sent X amount of bad data? Yes please.

SafePeer/Peerguardian-like filtering? Hell No.

If you want your BitTorrent client to act as your webserver, IRC client, RSS client, or firewall, just go back to Azureus.

I'm with this guy .

[v0ice]

I hear people saying they dont want the ip blocker, yet it just seems so repetative. Can you specify any reasons as to why you dont want it? The reason I was changing to this was to safe resources, installing another program does not accomplish that. You act as if this minor addon will change the whole program and make it 50% cpu. Do you know?

I understand you people don't want it, but can you at least state why so those of us who disagree can have something we can argue again'st.

The programmer wants to know the pros and cons of both sides, I'm assuming. If I'm wrong correct me.

-v0ice

ps. please refer to my first post in this topic if you want somewhere else to look for reason as why my ideas are negative.

[ScubaSteve]

as most ppl have stated in their posts disagreeing with it. why bother adding a feature to the program when theres Peerguardian and Protowall which both use drivers and can block the ips at kernel level, which is a far effective method and uses little to no resources. adding it is only going to waste development time and bloat the program with an unnescessary feature.

[YeCkeL]

God help the software project that decide feature sets by message board voting systems. In almost all cases the "Yes" votes will outnumber the "No" by a significant margin, no matter which feature it is.

Anyway, if you want to do a poll you should have been more clear about what kind of IP filter you mean. Is this for a PG/PW style massive blocklist, or simply an automatically created temporary filter to block IPs which are sending you bad data?

clearly you havent read the ip filter topic otherwise u would have known i was talking about the PG/PW type blocklist feature.

the blocking ip becoz of bad data feature should certainly be implemented. but the program doesnt need to have a blocklist feature like that of peerguardian/protowall. it is just going to bloat a perfectly good program.

Word i said this a while ago too.

[Ultima]

@v0ice: More programs doesn't mean more resources wasted. If you tried ProtoWall or PeerGuardian v2, you'd know that they waste close to 0% CPU time and something like 2-4MB RAM. Compare that with 100MB RAM (at least) with Azureus+Java and the massive CPU usage. Which one sounds worse now? =]

[grug]

I hear people saying they dont want the ip blocker, yet it just seems so repetative. Can you specify any reasons as to why you dont want it?

The first reason is that it is absolutely 100% pointless, and should not be a function of a torrent client. But regardless.

With more than a handful of IPs, application-level blocking (Peerguardian v1, Azureus Safepeer) sucks. To get decent performance, say that of ProtoWall or PeerGuardian v2, you need to block the IPs at the kernel level, installing an appropriate driver. This obviously goes beyond the scope of uTorrent's current development. It would require a significant amount of effort, development time and would be quite a task for the sole developer of uTorrent.

In addition, while application-level blocking sucks, it's an obvious choice when you only want to block one application's data. Since the application accepting the data is the same one blocking it, its a piece of cake.

Most people seem to be complaining that PeerGuardian blocks all your traffic, such as webbrowsers, yet they want its efficency in uTorrent.

Well its a choice between blocking all your data efficiently, blocking one applications data with huge bloat (Safepeer), or writing a kernel level driver that has to check what application the data is being sent to before deciding whether or not to block it. Which means reduced performance.

So if its non-bloated uTorrent-only IP blocking you want, you can't just plug some code in that checks the incoming IP against a list. Or you end up with a piece of shit like Safepeer. It would require essentially an entire seperate driver monitoring all the traffic that intended for uTorrent. This means a lot of development time, bugs, and no doubt unforeseen problems for uTorrent.

This is not meant to be an attack on uTorrent, but the first few versions of uTorrent had...shall we say, rather serious bugs that should not have made it to a release version. Based on its previous history, I would not like to see a buggy, poorly performing kernel-level IP blocker. On the other hand, I would not like to see a well tested kernel-level IP blocker that stole significant testing and development time from uTorrent's main purpose, that is, to download torrents. Nor would I like to see uTorrent shipped off with a halfassed poorly performing application level IP blocker.

And then there is the fundamental problem of processing that many incoming connections. You can either read directly through the blocklist like PeerGuardian v1 and have terrible performance and high CPU usage, or cache it in RAM. Bloat. Either way, you still have to parse through it hundreds/thousands/millions of times.

Regardless of my feelings of BitTorrent clients blocking IPs which I'm sure you're all aware of, I don't think it should be slated for version 1.2. There are a number of other features that should be part of uTorrent's core functionality before things like IP filtering blacklists are even thought about being implemented...Torrent queueing and super seeding for one.

Whether it's intended to have IP blocking eventually or not, I think the developer should stick with blocking IPs of those who send bad data and work on making the client solid at its sole purpose of downloading torrents before it branches off into other directions.

I really don't see why people can't just use Peerguardian 2 and tell it to not block HTTP traffic if they're really that paranoid and need to console themselves with a false sense of security.

If you tried ProtoWall or PeerGuardian v2, you'd know that they waste close to 0% CPU time and something like 2-4MB RAM. Compare that with 100MB RAM (at least) with Azureus+Java and the massive CPU usage. Which one sounds worse now? =]

2-4MB? Try 15-30MB.

(and FYI, just like azureus.exe only uses 2MB but javaw.exe uses >100MB, Protowall's application isn't indicative of its total memory usage either.)

I don't want uTorrent bloating up to 35MB in size.

And should we really be using Azureus as the benchmark?

[Ultima]

Hm that's weird... when I tried PG2 before, it didn't take that much memory... oh well, it was only limited testing anyhow (don't use IP filters)

[grug]

Well I just installed it 10 minutes ago. I don't actually use it. That was with just the P2P list btw, I didn't add any of the other filters.

[Ultima]

Bleh I don't know, maybe it's just memory playing games on me (don't remember when I last tested it -- could've sworn I had such results). Doesn't matter though, the point is that it uses far less resources in total than just Azureus alone =]

[winMX_67]

Ok even with HTTP block ON PG2 doesnt block browers. And yes PG2 uses like 2-4 mb and ive never seen CPU go over 2% from PG2!!

[grug]

Ok even with HTTP block ON PG2 doesnt block browers. And yes PG2 uses like 2-4 mb and ive never seen CPU go over 2% from PG2!!

I would like to know what mythical version of PG2 only uses 4MB RAM. Preferably accompanied with a Task Manager screenshot showing Mem + Peak Mem. It may idle at 2-4MB but when you get a large number of incoming connections you get large spikes.

Incidently, 4MB would more than double uTorrents current RAM usage on my PC, which is currently at 3,396KB.

uTorrent's peak usage is currently 5,888KB. PG2's peek usage is currently 43,296KB. Thats over 8 times the peak memory usage for an absolutely pointless feature.

[MP-A]

It wouldnt add any bloat. Pglite uses next to no resources, there should be a plugin made that would use the liteness of pglite with utorrent.

[v0ice]

It wouldnt add any bloat. Pglite uses next to no resources, there should be a plugin made that would use the liteness of pglite with utorrent.

This my feelings on the subject too. How do you people know it will cause bloat? Are you programing it?

thanks grug, for the good response.

[Ultima]

Since when did knowing when something is bloated require someone to program it?

Wikipedia article on Software Bloat"]Software bloat is a derogatory term used to describe the tendency of newer computer programs to use larger amounts of system resources (mass storage space, processing power and/or RAM) than older programs. It is also used in a more general context to describe programs which appear to be using more system resources than necessary, or implementing extraneous features. Software exhibiting these tendencies is referred to as bloatware or, less commonly, fatware.

IP range blocking (of the kind that PG2 or PW provide) is considered extraneous.

[winMX_67]

pg2.exe CPU 0 2,240 K and im running uTorrent.

[Ultima]

Not that I'm doubting you (it would confirm my experiences), but can you post a screenshot with pg2.exe in Task Manager? I'm only asking so that you can cast light on the doubt that people have of PG2 and its resource usage...

[ScubaSteve]

i had PG2 installed on here about 5 days ago whilst i helped someone use it and it used 0% cpu 6MB ram

if i install it again ill be sure to post screenshot of usage.

[Ultima]

What's more, if people are really afraid of even PG2's already low resource usage, they can use pglite. I haven't tested it myself, but supposedly, it requires even less resources (as it's a stripped, almost-GUI-less version of PG2 optimized "to consume as little CPU/RAM as possible"). Either way, a dedicated IP-range blocker is probably best left that way (dedicated) rather than adding the kitchen sink and coffemaker into utorrent.

[pike]

tip: use ProtoWall

[Niceguy]

I'd say yes to ip-filter functionality, it won't add much code and it won't use up any resources unless enabled. and given the current paranoia it might be an important point for people in switching to utorrent.