Timestamps are in GMT/BST.
[0:21] * Elly (n=elly@ool-457856bb.dyn.optonline.net) Quit (Read error: 110 (Connection timed out))
[0:24] * smains (n=smains@tor/session/x-1fa72692add2f051) has joined #freenet
[0:26] * Elly (i=elly@ool-457856bb.dyn.optonline.net) has joined #freenet
[1:32] * Romster (n=Romster@tor/session/x-690979fe9316af4e) Quit ("Connection reset by F.B.I.")
[2:04] * blank_ (n=smains@tor/session/x-85902fd48fa2bffc) has joined #freenet
[4:31] <nextgens> hi
[4:31] <nextgens> toad_: ping ?
[4:31] <nextgens> sandos: ping ?
[4:33] <nextgens> I need to talk to you about mailling issues on emu
[4:44] * sleon|sleep is now known as sleon|tuX
[4:50] * Hory (n=Miranda@82.78.27.85) has joined #FreeNet
[4:57] * nextgens is looking for sanity's gpg key
[4:57] <nextgens> and hopping it will be signed by toad
[5:01] <nextgens> I found it :)
[5:21] <nextgens> bbl
[5:49] * NullAcht15 (n=NullAcht@dslb-082-083-245-104.pools.arcor-ip.net) has joined #freenet
[5:57] * NullAcht15 (n=NullAcht@dslb-082-083-245-104.pools.arcor-ip.net) Quit (Read error: 104 (Connection reset by peer))
[7:15] * toad_ (i=toad@pdpc/supporter/active/toad-with-underline) Quit (Remote closed the connection)
[7:17] * toad_ (i=toad@pdpc/supporter/active/toad-with-underline) has joined #freenet
[7:18] * ChanServ sets mode +o toad_
[8:27] * greycat (i=rfc1413@wooledge.org) has joined #freenet
[8:39] * Romster (n=Romster@tor/session/x-fe9f20216e18b88d) has joined #freenet
[9:06] * FallingBuzzard (n=FallingB@66.151.22.70) has joined #freenet
[9:18] * NullAcht15 (n=NullAcht@dslb-082-083-245-104.pools.arcor-ip.net) has joined #freenet
[9:30] <toad_> hi folks
[9:36] <toad_> hmmm
[9:36] <toad_> are we fetching from a freeneturi, or from a clientkey...
[9:36] <toad_> hrrrrrrrm
[9:36] <toad_> well, if it's a freeneturi then it supports recursive archives
[9:36] <toad_> if not, it doesn't
[9:37] <toad_> well it could
[9:37] <toad_> hrrm
[9:38] <toad_> SSK@.../thing.tar.bz2//extra.tar//filename
[9:40] <toad_> hmmm
[9:40] <toad_> yeah
[9:40] <toad_> we have the FreenetURI
[9:41] <toad_> that corresponds to ONE freenet key, plus a list of meta-strings
[9:41] <toad_> so lets break it up at the top
[9:42] <toad_> do we want to allow redirects to actual URI's?
[9:42] <toad_> surely
[9:42] <toad_> including ones with lots of meta strings
[9:42] <NullAcht15> I don't like the stuff with the double slashes....
[9:42] <toad_> so
[9:42] <toad_> NullAcht15: you have a better idea?
[9:43] <NullAcht15> It looks unclean... hm, I'm thinking
[9:43] <toad_> NullAcht15: you think we should use single slashes?
[9:43] <NullAcht15> I mean, normally a double slash in a uri should have exactly the same meaning as a single one
[9:44] <toad_> that's not in the RFC as far as i know, it's just a unix convention
[9:46] <NullAcht15> besides, it's not obvious from the double slashes, what exactly you'll do with it. Do you try treating what comes before it as a manifest and extract the filename that comes after the double slash from the manifest and do a redirect
[9:46] <toad_> yes, basically
[9:46] <toad_> you treat what's before it as a manifest
[9:46] <NullAcht15> or do you treat what comes before the double slash as a zip file, unzip it and extract the file that way?
[9:46] <toad_> it might be a zip manifest or it might be a plain manifest
[9:46] <toad_> the metadata will tell you which
[9:47] <toad_> and we *might* treat zip's with no metadata as zip manifests with each file corresponding to itself with default mime type
[9:48] <toad_> NullAcht15: well, you have a better idea?
[9:48] <toad_> using single slash might work
[9:48] <toad_> a manifest can contain another manifest...
[9:48] <NullAcht15> what about something like tar:/(SSK@.../thing.tar.bz2/)/file
[9:48] <NullAcht15> not exactly like that
[9:49] <toad_> eeeeeewwwwww
[9:49] <NullAcht15> but I think you get the idea
[9:49] <toad_> brackets?!
[9:49] <NullAcht15> okay, the brackets are bad
[9:49] <toad_> yes, but they're essential for that kind of URL
[9:49] <NullAcht15> hm, maybe like this:
[9:50] <toad_> we can distinguish between zip manifest and ordinary manifest from the metadata
[9:50] <toad_> ordinary manifest is just metadata
[9:50] <NullAcht15> metadata:/tar:/SSK@.../thing.tar.bz2//manifest//filename
[9:50] <toad_> zip manifest would be metadata saying "read this url, it's a zip manifest"
[9:50] <FallingBuzzard> Shouldn't a ZIP file be treated just like a directory?
[9:50] <toad_> NullAcht15: but what if it's nested?
[9:50] <toad_> FallingBuzzard: hmm?
[9:51] <toad_> FallingBuzzard: zip files can contain subdirectories
[9:51] <FallingBuzzard> That is how Windows treats them
[9:51] <toad_> only windows XP
[9:51] <FallingBuzzard> You can change directory into a zip file
[9:51] <FallingBuzzard> and then into the subdirectories inside of it
[9:51] <toad_> we treat them as manifests; we have no concept of "directories", that's the basic problem
[9:51] <toad_> maybe we should do
[9:53] <toad_> well, the basic problem is that manifests and zip manifests can contain subdirs...
[9:54] <toad_> this doesn't necessarily have to be a big problem as long as the manifest itself has a name...
[9:54] <FallingBuzzard> If you treat them all like directories, it might be easier
[9:54] <toad_> well
[9:54] <FallingBuzzard> Then \ means the next directory down
[9:55] <toad_> what must be avoided at all costs is ambiguity - is x/y inside the zip called x, or is it directly in the manifest as x/y ?
[9:55] <toad_> ummm
[9:55] <toad_> /
[9:55] <toad_> URIs, not windows crappy pathnames :)
[9:55] <FallingBuzzard> If x is a zip then it is y inside of that zip
[9:55] <toad_> right, that means we have a concept of directories
[9:56] <toad_> .... which doesn't necessarily mean we can't have full paths in both zips and manifests
[9:56] <toad_> so you end up with a manifest:
[9:56] <toad_> index.html
[9:56] <toad_> images/baah.png
[9:56] <toad_> text - ZIP file
[9:57] <toad_> text/blah - INVALID, ignore and log an error
[9:57] <toad_> messy
[9:57] <toad_> especially if we get text/blah before we get text !
[9:57] <toad_> but unavoidable as zip's can contain pathnames
[9:58] <FallingBuzzard> is it too messy to require a zip file to end in .ZIP?
[9:58] <toad_> unless we don't want people inserting zips directly
[9:58] <toad_> far too messy imho yes, and anyway we have a content-type
[9:58] <toad_> only windows uses filenames to determine content type
[9:59] <FallingBuzzard> SSK@blablabla/zipfile.zip/filename.html
[9:59] <toad_> ugh
[9:59] <toad_> horrible
[9:59] <FallingBuzzard> SSK@blablabla/tarfile.tar/filename.html
[9:59] <toad_> still horrible!
[9:59] <toad_> and anyway it'd be tar.gz
[10:00] <FallingBuzzard> Thats right
[10:00] <toad_> SSK@..../CofE/images.tar/evil.png
[10:00] <FallingBuzzard> yes
[10:00] <NullAcht15> why? You could just say, if you have a pathname x/y, you'll first have to fetch x, determine what it is and then use x to determine how to get at y....
[10:00] <toad_> anyway we can't use tarfiles etc directly, we still need metadata...
[10:00] <FallingBuzzard> Because they could be nested archives
[10:01] <toad_> NullAcht15: that means having directories; specifically, it means that neither zips nor manifests can have filenames with slashes in them
[10:01] <NullAcht15> not necessarily
[10:01] <toad_> whereas right now, we can use existing zip's as manifests, without making any requirements on them, and manifests can have slashes in filenames
[10:02] <NullAcht15> it might mean that if they have names with slashes, these names implicitly specify subdirs
[10:02] <toad_> invisible subdirs
[10:03] <toad_> which might have a name clash with a sub-manifest or sub-zip
[10:03] <toad_> of course we CAN detect such things
[10:03] <NullAcht15> example: SSK@.../manifest/container.zip/image.png -> you first fetch SSK@.../manifest to determine how to get at container.zip. Fetch that to determine how to get at image.png
[10:03] <toad_> what if manifest contains a redirect to a file called container.zip/image.png?
[10:03] <NullAcht15> yes, nameclashes.... hm...
[10:05] * toad_ thinks maybe the whole idea of zip manifests is upside down... maybe we should have a special kind of redirect that goes inside a zip file, and then have the metadata a conventional manifest
[10:05] <toad_> which could itself be stored inside one, via multi-level (or rather indirect) metadata
[10:05] <NullAcht15> hm, that might mean requiring that neither manifests nor containers (nor anything else that could be treated as a directory) can only specify "shallow" dirs
[10:06] <toad_> NullAcht15: huh?
[10:06] <toad_> NullAcht15: ah yeah
[10:06] <NullAcht15> meaning filenames within manifests can't contain / and zips used as containers can't have subdirs. Why not?
[10:06] <toad_> NullAcht15: but we want to be able to use existing archives as containers
[10:06] <toad_> e.g. the linux source tarball
[10:07] <NullAcht15> Is that really necessary?
[10:07] <toad_> what?
[10:07] <toad_> using existing archives as containers isn't absolutely necessary
[10:07] <toad_> but it would be very nice
[10:08] <NullAcht15> well, it comes down to having to define a clean way for defining directory structures...
[10:08] <toad_> if we just consider metadata manifests...
[10:08] <toad_> then we would have to have shallow directories
[10:08] <toad_> but that's easy enough with metadata manifests
[10:09] <toad_> of course if we just say no archive recursion...
[10:09] <toad_> that would solve the problem
[10:10] <toad_> no expanding archives inside archives
[10:10] <NullAcht15> as in no "ln -s bla blubb; ln -s blubb bla"....
[10:10] <toad_> NullAcht15: no, not talking about circular here
[10:10] <toad_> related issue
[10:11] <NullAcht15> okay, but that's another thing that would need to be detected
[10:11] <toad_> should we allow blah.zip//filename where blah.zip is just an ordinary zip file?
[10:11] <toad_> or do we require metadata pointing to it?
[10:11] <NullAcht15> I still think we should get rid of double slashes
[10:12] <toad_> if so, we can do what i suggested above - have a special redirect which tunnels inside a zip file?
[10:12] <NullAcht15> uh, let me try to understand that...
[10:13] <toad_> it would waste a bit of metadata space, but not a lot after we compress the metadata
[10:13] <FallingBuzzard> I don't know enough about the metadata manifest to have any suggestions
[10:13] <toad_> FallingBuzzard: what do you need to know about it? :)
[10:13] <toad_> metadata manifest is just a list of filenames and redirects
[10:14] <FallingBuzzard> Does it include directories?
[10:14] <toad_> except they don't have to be redirects, they can be splitfiles, manifests, or whatever
[10:14] <NullAcht15> Why not just say that if you have a pathname a/b/c/d, you'll have to first fetch a, use taht to determine how to fetch b, use that to determine how to fetch c.....
[10:14] <toad_> FallingBuzzard: the filenames at present can have slashes in them
[10:14] <FallingBuzzard> They can't in normal URL's.
[10:14] <NullAcht15> toad_: And I think we should get rid of that
[10:14] <toad_> NullAcht15: that works well with metadata manifests
[10:15] <toad_> NullAcht15: we COULD use that in archives too
[10:15] <NullAcht15> I mean, that _is_ how you normally define a directory tree...
[10:15] <toad_> NullAcht15: indeed
[10:15] <NullAcht15> you have a chain of redirects
[10:15] <NullAcht15> one redirect at every /
[10:16] <NullAcht15> no need for double slashes
[10:16] <toad_> well, how important is it for people to be able to do CHK@..../linux-2.6.89.tar.bz2/include/widget.h ?
[10:17] <NullAcht15> you'd need to download linux-2.6.89.tar.bz2 completely anyway
[10:17] <NullAcht15> maybe we should even define a new archive format that specifically does not support subdirs
[10:18] <toad_> well, or CHK@...../freenet-ext.jar/Thingy.class
[10:18] <toad_> or CHK@..../freenet-20060708.tgz/freenet.jar
[10:18] <NullAcht15> toad_: the last one isn't even a problem
[10:18] <NullAcht15> the last two
[10:18] <toad_> true
[10:18] <toad_> CHK@..../freenet-20060708.tgz/lib/freenet.jar
[10:18] <toad_> :)
[10:19] <NullAcht15> even CHK@..../freenet-20060708.tgz/freenet.jar/Thingy.class would not be a problem
[10:19] <toad_> the question is, do we let people access archives directly, or do we force them to make a metadata redirect?
[10:19] <NullAcht15> CHK@..../freenet-20060708.tgz/lib/freenet.jar would not be a problem if lib were a manifest file
[10:19] <toad_> the latter could perhaps solve some of these problems
[10:19] <toad_> well...
[10:19] <toad_> we can treat it as such
[10:20] <toad_> a zip file will not permit a directory and a file to have the same name, will it?
[10:20] <NullAcht15> I don't think so
[10:20] <toad_> my root doesn't seem to
[10:21] <toad_> ok
[10:21] <toad_> so
[10:21] <NullAcht15> I'd say if we are told to fetch CHK@..../freenet-20060708.tgz/file for example, and we find that feenet-20060708.tgz contains any subdirs at all, the request should be rejected
[10:21] <toad_> well, we could do that
[10:21] <toad_> or we could just build the directory tree within it
[10:22] <NullAcht15> which, imo, would be unclean
[10:22] <NullAcht15> or would it?
[10:22] <NullAcht15> could that lead to name clashes?
[10:23] <toad_> we could even do that with ordinary manifests... if we build an actual tree structure, it's reasonably easy to check for duplicates
[10:23] <NullAcht15> not if manifests are required not to include subdirs...
[10:23] <toad_> it could, but if we handle it right, we can detect them easily enough
[10:24] <toad_> it would mean we have to keep the whole directory tree in RAM for any archives...
[10:24] <NullAcht15> using metadata, container names won't even have to end in zip or tar.gz
[10:24] <NullAcht15> oh, and another thing
[10:25] <toad_> what we'd have to do ...
[10:25] <toad_> SSK@..../CofE/images/blah.png
[10:25] <toad_> we fetch SSK@.../CofE
[10:25] <NullAcht15> I don't think manifests should be completely defined in metadata. The metadata should only say "this is a manifest, treat it as such"
[10:26] <toad_> that's a tar.bz2 archive
[10:26] <NullAcht15> the actual redirects should be in the data part
[10:26] <toad_> we then look for images/blah.png
[10:26] <toad_> we don't find it
[10:26] <toad_> so we look for images
[10:26] <toad_> we find it
[10:26] <toad_> we assume the archive is sane, or we verify its integrity on unpacking it in the first place
[10:27] <toad_> hmmm
[10:27] <toad_> NullAcht15: I don't get it
[10:27] <toad_> NullAcht15: yo m
[10:27] <NullAcht15> what?
[10:27] <toad_> NullAcht15: you mean in HTML, or what?
[10:27] <NullAcht15> no
[10:27] <toad_> well
[10:27] <toad_> the current plan is
[10:27] <toad_> a manifest is a kind of metadata document
[10:27] <toad_> which defines a mapping between filename and another metadata document
[10:27] <toad_> which is then parsed
[10:27] <NullAcht15> about the same format as used now. But not in the metadata space, but in the actual data part
[10:28] <toad_> NullAcht15: but it's metadata! data is what is delivered to the user
[10:28] <NullAcht15> well, why should the user not be able to directly download, say, SSK@..../CofE/images, where images is a manifest
[10:29] <NullAcht15> he /can/ already dl SSK@..../CofE, unless you want to put the whole tar.bz2 archive into metadata space
[10:29] <toad_> he should be able to, if it's a zip
[10:29] <toad_> if it's a manifest, he can download the default document
[10:29] <toad_> that's the problem - we may, and usually do, have a default document
[10:30] <toad_> so if he goes to SSK@.../CofE, he gets the default document, not the archive itself
[10:30] <NullAcht15> no, the default document should only be delivered for SSK@..../CofE/images/ not for SK@..../CofE/images
[10:30] <NullAcht15> not the trailing slash
[10:30] <NullAcht15> note
[10:30] <toad_> isn't that user unfriendly? on the general WWW, if you go to .../images/ you get the same result as .../images
[10:30] <NullAcht15> no
[10:31] <NullAcht15> sometimes not
[10:31] <NullAcht15> btw, this way, by requesting SSK@..../CofE/images instead of SSK@..../CofE/images/, you could easily get a sort of directory listing
[10:32] <NullAcht15> which would otherwise be impossible
[10:33] <toad_> hmmm
[10:33] <NullAcht15> I mean, if I request SSK@..../CofE/images, I want SSK@..../CofE/images, not SSK@..../CofE/images/index.html or something like that
[10:34] <toad_> if i request http://www.microsoft.com
[10:34] <NullAcht15> maybe fproxy could decide to handle this case differently - but fcpv2 shouldn't
[10:34] <toad_> i want http://www.microsoft.com/index.asp
[10:35] <NullAcht15> on the normal web, people have good reasons to not let you have dir listings whereever you wnat the,
[10:35] <NullAcht15> them
[10:35] <toad_> ... or whenever you are sloppy with a url...
[10:36] <toad_> which largely isn't a problem on freenet
[10:36] <toad_> since urls aren't guessable anyway
[10:36] <NullAcht15> yepp
[10:37] <toad_> so what are you saying? if we reach a manifest, and it doesn't define a default document, and there are no more metastrings left in the url, we return a generated index?
[10:37] <NullAcht15> depends on what was requested
[10:38] <NullAcht15> if a/b/c was requested, and b doesn't tell us where to find c, return an error
[10:38] <NullAcht15> if a/b/ was requested and b defines c as the default doc, return c
[10:38] <NullAcht15> if a/b was requested, return b
[10:39] <NullAcht15> without looking at it
[10:39] <toad_> but what if b is a metadata document?
[10:39] <toad_> it wouldn't be very readable
[10:40] <toad_> and generally speaking should not be returned to a user
[10:40] <NullAcht15> if a/b/ was requested and b doesn't define a default doc, return an error
[10:40] <toad_> (metadata is binary)
[10:40] <toad_> how about this:
[10:40] <NullAcht15> toad_: so what? if the user wants the metadata doc, give him that
[10:41] <toad_> NullAcht15: that encourages abusive use of metadata, I object to that; metadata should only be visible if you explicitly ask for it
[10:41] <toad_> how about this:
[10:41] <NullAcht15> As I said, maybe fproxy should catch this case, so normale freenet users won't ever see such a thing
[10:41] <toad_> NullAcht15: more bloat in fproxy. bad thing. fproxy should be a very lightweight layer on top of FCP
[10:41] <NullAcht15> but it should only be for "surfing"
[10:42] <NullAcht15> toad_: Are you going to implement the anonymity filter in fproxy or below it?
[10:43] <toad_> how about this: both archive manifests and metadata manifests are allowed to contain slashes. they are required to be consistent - no two entries with the same filename, no directory with the same filename as a file. so when we want to fetch x/y/z/a, we check x, then we check x/y, then we check x/y/z, then we check x/y/z/a
[10:43] <toad_> NullAcht15: below it, as an option on FCP requests
[10:43] <NullAcht15> hm, anonymity filter is highly application specific
[10:44] <toad_> NullAcht15: well, if you implement a plugin for a web browser to do FCP, you would still want it
[10:44] <toad_> if we are talking about a transport plugin, rather than major changes across the browser
[10:44] <toad_> which is what some people seem to assume
[10:45] <NullAcht15> a freenet mail application, for example, would have much different needs from an anonymity filter than fproxy or something like it
[10:45] <toad_> indeed, so it wouldn't ask for filtering
[10:45] <toad_> well actually if people were sending HTML mails i suggest that it would have identical needs
[10:45] <NullAcht15> hm...
[10:46] <NullAcht15> but it would be impractical at the fcp level anyway. HTML mail are send as attachments to normal mails. the filter would need to detect that first
[10:47] <toad_> well it depends on your implementation, but if there is demand for it we can have a way to access the filter on submitted content via FCP
[10:48] <toad_> i just don't want to have 12 different filters out there, all of which have numerous well known (to the others!) vulnerabilities
[10:48] <NullAcht15> anyway, I can't think of many more applications besides browsers who'd need a filter anyway...
[10:49] <NullAcht15> most others would be written specifically for freenet anyway
[10:49] <toad_> don't you agree that a proliferation of filters is probably a bad thing?
[10:49] <NullAcht15> Yes, but I also don't think it would happen
[10:50] <toad_> generally speaking, one of the goals of 0.7 is to cut the fproxy-only functionality to an absolute minimum
[10:51] <NullAcht15> okay, good idea in general, but I don't think that should mean moving functionality that is only useful for fproxy anyway into fcp
[10:51] <toad_> it might be useful to somebody someday writing a browser plugin
[10:52] <NullAcht15> This sort of problem is often solved quite successfully with shared objects/libraries
[10:52] <toad_> not in the java world; we provided libraries for FCP, and nobody uses them
[10:53] <NullAcht15> maybe that's because hardly anyone ever heard of them?
[10:54] <toad_> well, what about a mozilla plugin?
[10:54] <toad_> i think we can be 99% confident that it would not be written in java
[10:55] <toad_> and therefore that it would be a PITA to link it with a java library
[10:55] <NullAcht15> Can't Java use C-libraries?
[10:55] <NullAcht15> pretty much anything can use c libraries
[10:55] <toad_> yes, but we can't bundle them unless we also provide a java alternative with identical functionality
[10:56] <NullAcht15> which would be doubled effort and twice the potential for errors...
[10:56] <toad_> right, so we only ship pure java
[10:57] <NullAcht15> Why can't you bundle them without also supplying a java alternative? Freenet has no support to speak of for anything besides linux/x86 and windows anyway
[10:58] <toad_> only because java only supports linux/x86 and windows
[10:58] <NullAcht15> so?
[10:58] <toad_> freenet will run fine on mac, sparc, anywhere with a solid JVM
[10:58] <toad_> once we run on GCJ we will more or less run anywhere
[10:59] * jieryn (n=jesse@unaffiliated/jieryn) has joined #freenet
[10:59] <NullAcht15> You could do that already if you didn't choose java#
[10:59] <toad_> sure
[10:59] <NullAcht15> okay, I'm about to start a language flamewar, so I#ll stop now
[11:00] <NullAcht15> argh, it's 5 pm already, I need to do more important things...
[11:00] <NullAcht15> more important for me, anyways...
[11:00] <toad_> :)
[11:00] <toad_> i'm posting to the list
[11:00] <toad_> a suggestion
[11:01] <NullAcht15> which one?
[11:01] <NullAcht15> which list, I mean?
[11:01] <toad_> it's on tech and devl now, read it :)
[11:01] <toad_> Subject: [freenet-dev] Double slashes considered harmful?
[11:02] <toad_> wait a minute... the search is only necessary if we want to support zip manifests with no internal metadata
[11:03] <toad_> !
[11:04] <toad_> okay, so all we have to do is require that there be metadata, even in an archive manifest... an archive manifest must contain a metadata manifest, referring to stuff inside the archive
[11:05] <toad_> and if we are going to do it that way we might as well go the whole hog
[11:05] <toad_> => only support archive manifests via ARCHIVE_REDIRECT
[11:05] <toad_> which simply contains a redirect to the archive, plus the name of the file within it that we want to extract
[11:06] * Elly (i=elly@ool-457856bb.dyn.optonline.net) Quit (Connection timed out)
[11:11] <toad_> and the reply to that, which is also from me
[11:16] * linyos (n=linyos@109-17.73-24.tampabay.res.rr.com) has joined #freenet
[11:26] <toad_> hi linyos
[11:34] * NullAcht15 (n=NullAcht@dslb-082-083-245-104.pools.arcor-ip.net) Quit (Remote closed the connection)
[12:05] * rah (n=rah@cpc1-nott8-4-1-cust246.nott.cable.ntl.com) Quit (Read error: 110 (Connection timed out))
[12:42] * Romster (n=Romster@tor/session/x-fe9f20216e18b88d) Quit (SendQ exceeded)
[12:45] * Elly (i=elly@ool-457856bb.dyn.optonline.net) has joined #freenet
[12:51] * NullAcht15 (n=NullAcht@dslb-082-083-245-104.pools.arcor-ip.net) has joined #freenet
[13:02] * NullAcht15 changes topic to 'http://freenetproject.org | Stable: Upgrade to 5105 | Unstable:60266 | Channel logs: http://sbc.lir.dk/LogBot/ or http://newton.matcmp.ncc.edu/~lockej/freenet/chanlog/ | #freenet-alphatest for anyone who wants to help test 0.7 | DO NOT use the freenet packages supplied by debian! | FAQ: http://freenetproject.org/wiki/FaqsAndHowtos | A fresh install of freenet will take some time to become useful, be patient'
[13:28] * jieryn (n=jesse@unaffiliated/jieryn) has left #freenet
[13:34] * Ribs (n=ribs@riblet.plus.com) has joined #freenet
[13:43] <toad_> hey folks
[13:43] * toad_ back
[13:45] <TheSeeker> are your 0.7 nodes up toad?
[13:46] <toad_> no
[13:46] <toad_> you would like me to bring them up?
[13:47] <TheSeeker> If you aren't testing anything right now, I guess there's no reason to.
[13:57] * Ribs (n=ribs@riblet.plus.com) Quit (Read error: 104 (Connection reset by peer))
[14:05] <toad_> Subject: [Tech] Manifests, double slashes etc
[14:55] <NullAcht15> I like it. Sounds like a good, clean way to build arbitrary directory hierarchies
[15:53] * unknown_ (n=NullAcht@dslb-082-083-226-129.pools.arcor-ip.net) has joined #freenet
[15:57] * NullAcht15 (n=NullAcht@dslb-082-083-245-104.pools.arcor-ip.net) Quit (Nick collision from services.)
[15:58] * unknown_ is now known as NullAcht15
[16:04] <linyos> booyaa
[16:16] <toad_> hi linyos
[16:16] * toad_ reads the bit on 2nd ren about freenet and porn... looks interesting, maybe some day i read the rest of it
[16:16] <greycat> I've read bits and pieces of the 2nd renaissance sites. they're enormous....
[16:17] <toad_> just one big rant then?
[16:17] <greycat> no, a whole bunch of little rants all linked together
[16:17] <toad_> ahhh
[16:17] <toad_> :)
[16:18] <toad_> the fact that it mentions freenet strongly suggests that it's original freenet content :)
[16:18] <toad_> therefore a good thing no matter how large and whacky :)
[16:18] <toad_> greycat: any opinions on eliminating double slashes in freenet 0.7 URLs?
[16:19] <greycat> I haven't been following it, but making freenet work with older versions of wget is surely a nice thing
[16:20] <greycat> I think newer ones finally fixed the // issue, but I haven't confirmed it
[16:20] <toad_> older versions? they finally fixed it?
[16:20] * toad_ checks
[16:20] <toad_> nope
[16:20] <toad_> not in my copy
[16:20] <toad_> $ wget http://127.0.0.1:8888/SSK@MTI5V9WVOphqFv7raeyVBSIcd4IPAgM/2Rtext1/1//
[16:20] <toad_> --21:23:26-- http://127.0.0.1:8888/SSK@MTI5V9WVOphqFv7raeyVBSIcd4IPAgM/2Rtext1/1/
[16:20] <toad_> => `index.html.1'
[16:21] <linyos> most users don't know even understand the difference between a url and a google search query.
[16:21] <greycat> wget (1.10-1) unstable; urgency=low
[16:21] <greycat> * double slash problem fixed
[16:21] <greycat> (closes: Bug#184415)
[16:21] <toad_> greycat: yay
[16:21] <linyos> don't even understand...
[16:22] <toad_> linyos: true, except you type search queries into a different box
[16:24] <linyos> i think a lot of people just type them into google.
[16:24] <linyos> the uris, i mean.
[16:24] <toad_> lol, that even works
[16:25] <greycat> griffon:/stuff$ wget http://127.0.0.1:8888/SSK@rBjVda8pC-Kq04jUurIAb8IzAGcPAgM/TFE//thelist.html
[16:25] <greycat> 16:28:23 (82.53 KB/s) - `thelist.html' saved [456079]
[16:25] <toad_> greycat: wooh, cool
[16:25] <toad_> i still like one path component = one manifest (directory) though
[16:27] <linyos> it would be neat to have an AI that learns a user's browsing habits (vis-a-vis updated sites) and keeps them fresh in the cache automatically.
[16:28] <toad_> you don't really need an AI for that
[16:28] <toad_> just some heuristics
[16:28] <toad_> "subscribe to any site the user has visited twice in the last month"
[16:28] <linyos> sure, whatever you call it.
[16:29] <toad_> YOU can code it, when i write fproxy! :)
[16:29] <linyos> ok, it's a deal. i'll code that.
[16:29] <toad_> fproxy will maybe be a separate code module
[16:29] <toad_> to try to not put people off too much
[16:30] <toad_> with a simple interface to the main code, and the ability to work over FCP
[16:30] <toad_> well, separating it is an idea that's been discussed...
[16:31] <toad_> certainly we should try to keep it separable anyway
[16:31] <toad_> linyos, greycat , do you have any opinions on DBRs? in particular, can the interval be just a fixed number of milliseconds, or would it be useful to support e.g. calendar months?
[16:32] <greycat> Well... are you planning to use Freenet just on earth, or in outer space too? :)
[16:32] <linyos> hmm, tricky.
[16:33] <greycat> I'd think that using fixed number of {milli,}seconds should be sufficient.
[16:33] <linyos> greycat: but then it drifts out of sync with the calendar
[16:33] <greycat> people may want to make weekly sites, but anything beyond that, and you really need to use edition numbers instead
[16:34] <linyos> i mean, you might want to update it at four in the morning every day.
[16:35] <linyos> you might make it a "times per day" value.
[16:36] <linyos> ie, "4/1" means "four times per day" and "1/2" means "every two days"
[16:36] <greycat> how were you planning to store 1/7?
[16:36] <linyos> like, as a "1" and a "7"
[16:36] <linyos> meaning once every seven days.
[16:37] <greycat> not as a floating point value, then. OK.
[16:37] <linyos> yeah, an integer ratio
[16:37] <toad_> ummm
[16:37] <toad_> you store it as an interval
[16:37] <toad_> right/
[16:37] <toad_> ?
[16:38] <toad_> four times per day = once every 6 hours
[16:38] <toad_> no?
[16:38] <FallingBuzzard> How about 1 and 7D keeping the letter D for future expansion
[16:38] <FallingBuzzard> Think W, M, Y
[16:38] <toad_> the whole point with DBRs is you can go from the redirect and today's date to the expected URI, you don't have to search around
[16:39] <toad_> well, we used to use a YYYYMMDDHHMMSS format
[16:39] <toad_> in 0.3
[16:39] <toad_> obviously i'd prefer to do it more efficiently than a text string as above, but that seems like a reasonable basic principle
[16:40] <toad_> the alternative is the 0.5 way, which is just have an interval in seconds
[16:42] <toad_> or we could just scrap DBRs and implement TUKs straight off :)
[16:42] <FallingBuzzard> DBR's are an ugly hack
[16:42] <toad_> but it would be nice to have a proper alpha test this year :)
[16:44] <linyos> aren't there leap seconds and weird stuff like that?
[16:45] <FallingBuzzard> Leap years
[16:45] <linyos> i thought the earth was slowing down or something
[16:45] * FallingBuzzard looks at his personal atomic clock
[16:48] * greycat (i=rfc1413@wooledge.org) Quit ("This time the bullet cold rocked ya / A yellow ribbon instead of a swastika")
[16:48] <FallingBuzzard> Start time in YYYYMMDDHHMMSS format with an offset value in the format 00000001000000 meaning updated every day
[16:48] <FallingBuzzard> 00000002000000 is updated every 2 days
[16:49] * linyos looked it up.
[16:49] <linyos> they insert an extra second every few years or so
[16:50] <linyos> meaning that some days are 86401 seconds long.
[16:51] <linyos> so either way, it won't drift much.
[16:52] <FallingBuzzard> Interesting http://tycho.usno.navy.mil/leapsec.html A leap second is added about 1 time a year since 1972
[16:55] <linyos> When a leap second occurs, so that the UTC day is not exactly 86400 s long, a discontinuity occurs in the Unix time number. The Unix time number increases by exactly 86400 each day, regardless of how long the day is.
[16:55] <linyos> so some unix-logical "seconds" are actually two real seconds long.
[16:56] <linyos> meaning thus is totally irrelevant to us.
[16:56] <linyos> this...
[16:56] * FallingBuzzard (n=FallingB@66.151.22.70) has left #freenet
[17:07] <nextgens> hi
[17:13] * blank_ (n=smains@tor/session/x-85902fd48fa2bffc) Quit ("Leaving")
[17:13] * Sugadude (n=Sugadude@tor/session/x-512a54ebc99bc1e7) has joined #freenet
[17:38] <nextgens> toad_: yes, you can distribute the pubkey of the CA
[17:45] * unknown_ (n=NullAcht@dslb-082-083-211-154.pools.arcor-ip.net) has joined #freenet
[17:46] * unknown_ (n=NullAcht@dslb-082-083-211-154.pools.arcor-ip.net) Quit (Client Quit)
[17:50] * unknown_ (n=NullAcht@dslb-082-083-211-154.pools.arcor-ip.net) has joined #freenet
[17:52] * rah (n=rah@cpc1-nott8-4-1-cust246.nott.cable.ntl.com) has joined #freenet
[18:01] * NullAcht15 (n=NullAcht@dslb-082-083-226-129.pools.arcor-ip.net) Quit (Connection timed out)
[18:09] <toad_> hi folks
[18:10] * sleon|tuX (i=test@e180032049.adsl.alicedsl.de) Quit (Read error: 110 (Connection timed out))
[18:15] <nextgens> hi toad_
[18:18] <hobx> my fp.org address does not work
[18:23] <toad_> how were you accessing it?
[18:27] <hobx> no
[18:27] * sleon|tuX (i=test@e180012062.adsl.alicedsl.de) has joined #freenet
[18:27] <hobx> it is just a forward
[18:27] <hobx> do you have root?
[18:30] * smains (n=smains@tor/session/x-1fa72692add2f051) Quit (Success)
[18:31] <hobx> somebody needs to remove the alias which sends oskar -> hobbex because there is no hobbex user on the new machine
[18:37] * unknown_ (n=NullAcht@dslb-082-083-211-154.pools.arcor-ip.net) Quit (Read error: 104 (Connection reset by peer))
[18:38] <hobx> my user on "emu" is "oskar" so there is no need for an alias to forward oskar to hobbex
[18:38] <hobx> (line 47 in /etc/aliases)
[18:42] <nextgens> hobx: ok, I'll do
[18:43] <nextgens> done
[18:44] <nextgens> and postfix has been updated
[18:58] * Ash-Fox (i=UNKNOWN@eco146.neoplus.adsl.tpnet.pl) Quit (Nick collision from services.)
[18:58] * Ash-San (i=UNKNOWN@ech97.neoplus.adsl.tpnet.pl) has joined #FreeNET
[19:00] <hobx> it still doesn't work?
[19:00] <hobx> :-(
[19:04] * Ash-San is now known as Ash-Fox
[19:09] <nextgens> hobx: could you explain me what to do please ? :)
[19:10] <nextgens> the line has been removed and I did postalias /etc/aliases
[19:10] <nextgens> it should be sufficient, isn't it ?
[19:10] <nextgens> what's wrong with your account ?
[19:14] <nextgens> hobx: do you want me to put your .forward in the aliases file ?
[19:15] <hobx> well, no something came through
[19:15] <hobx> perhaps it just very slow
[19:16] <hobx> thanks for helping
[19:16] <hobx> above should have said "now something came through"
[19:16] <nextgens> np
[19:44] * Hory (n=Miranda@82.78.27.85) Quit ("http://CyberLore.net - Recommendations on the best games, freeware and websites.")
[22:02] * Sugadude (n=Sugadude@tor/session/x-512a54ebc99bc1e7) Quit ("Zzz")
[22:08] * linyos (n=linyos@109-17.73-24.tampabay.res.rr.com) Quit (Read error: 110 (Connection timed out))
[22:32] * sleon|tuX (i=test@e180012062.adsl.alicedsl.de) Quit (Remote closed the connection)
[22:56] * sleon|tuX (i=test@e180012062.adsl.alicedsl.de) has joined #freenet
These logs were automatically created by Jay Oliveri with his gimp hapi on irc.freenode.net.