You can subscribe to this list here.
2000 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
(9) |
---|---|---|---|---|---|---|---|---|---|---|---|---|
2001 |
Jan
(46) |
Feb
(37) |
Mar
(8) |
Apr
(76) |
May
(63) |
Jun
(9) |
Jul
(18) |
Aug
(7) |
Sep
(8) |
Oct
|
Nov
|
Dec
(7) |
2002 |
Jan
(1) |
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
From: Kirk R. <ki...@br...> - 2001-02-23 16:29:42
|
Hi again folks: Well, I looked through the archives for January and February but didn't find any discussion of the mail or smtp specs. I saw references to the smtp modules in cvs which I haven't looked for yet. I couldn't manage to find a December archive though. I don't find the geocrawler linking convention intuitive either. I found the 2001 arives available at lists/3/SourceForge/8627/2001 but there didn't seem to be an entry for 2000 at least under the lists/3/SourceForge/8627/2000 structure. I tried to do a couple of searches with the search feature but it seems useless. It couldn't even find anything with mail. I have take a preliminary look through the cvs mail module though. It appears that all of the freenet specific stuff is in download.java, is that correct? Also reading through the other messages seemed to indicate that SVKs are actually not updatable yet. Is that true? I have seen a number of references to freenetlib and libfreenet. Are these separate packages? I looked through the freenetlib directory in cvs and it appears to be python. Yet Mark and Brandon made reference to a c library interface. Where might I find this library? Kirk |
From: Kirk R. <ki...@br...> - 2001-02-23 13:25:38
|
Okay, I'll check that out. Now I know this is going to sound really silly but I went to http://lists.sourceforge.net/mailman/listinfo/eof-dev/. I mean the mailing list is on sourceforge right? This is getting to be something like "the plans are on display at the bottom of a filing cabinet, behind a locked door, in the basement of the public planning committee building, guarded by a tiger." Kirk -- Kirk Reiser The Computer Braille Facility e-mail: ki...@br... University of Western Ontario phone: (519) 661-3061 |
From: Brandon <bl...@ut...> - 2001-02-22 22:17:45
|
> Well, the mail archive has one month listed, December with a total of > one article only mentioning that you put freenetmail in cvs. It would > appear that it won't post January's archive until the end of the > month. No wait a minute, this is almost the end of February, okay > I'll play, where have the archives gone? I don't know what where you went, but here's a link to archives including the message you just posted asking where the archives are: http://www.geocrawler.com/lists/3/SourceForge/8627/0 |
From: Kirk R. <ki...@br...> - 2001-02-22 15:16:57
|
Well, the mail archive has one month listed, December with a total of one article only mentioning that you put freenetmail in cvs. It would appear that it won't post January's archive until the end of the month. No wait a minute, this is almost the end of February, okay I'll play, where have the archives gone? Kirk |
From: Kirk R. <ki...@br...> - 2001-02-22 14:58:21
|
I will read the archives. I thought the new specs were going to give us a nice clean English interface to the server? After looking them over it seems very straight forward. Kirk -- Kirk Reiser The Computer Braille Facility e-mail: ki...@br... University of Western Ontario phone: (519) 661-3061 |
From: Brandon <bl...@ut...> - 2001-02-22 04:18:56
|
> Hi Folks: Well I was told that this is the place to be to discuss > mail specifications for the freenet. I have downloaded the current > cvs but there appears to be no documentation of how the current code > works. Would someone mind giving a description of the protocal for > the current implementation? The documentation is in the mailing list archives. The mailing list archives are short, so I don't feel bad telling you to go read them. Then come back and I'l answer any questions because I may not have been clear. I'll take any clarifications I need to make and put them in an f-mail FAQ. > I will attempt to look through the code but have to admit I don't > really know java. I was hoping for something which was written in c. It's imperative to write everything in Java so that the library we use can keep up with the rapidly changing Freenet protocol. If you want to implement it in C that would be cool. You can use libfreenet, the Freenet C client library. It should be pretty easy to write in C and libfreenet is expected to not lag *too* far behind the reference implementation. |
From: Kirk R. <ki...@br...> - 2001-02-22 03:36:16
|
"Mark J. Roberts" <mj...@st...> writes: > Hey! The CVS WORKS! Yay for EOF! Haven't I seen you someplace before? I think things is gettin' a bit incestuous around here. > Dunno. Now I get myself all juiced up and excited about getting some answers and yu don't know! > We could certainly implement the protocols in C. Well, I will volunteer to do just that should I ever get an understanding about what we're supposed to be doing. Let me throw out a few of my ideas and you may all laugh at me. I was tinkin' see, that a person could decide on a KSK which would be their mail address. This KSK would point to a public pgp key. They would also have a second KSK which pointed to a sequence number. I think the KSK, let's call him "Joe Sixpack" would actually point to an svk containing the public pgp key. "Joe Sixpack-Seq" points to another svk containing an eight hex digit sequence number. So I want to send a note to Joe, I might insert it to "Joe Sixpack-0000000000000001". Joe's mail daemon is the keeper of the "Joe Sixpack-Seq" key so it updates that each time a message is retrieved with a higher number like 0000000000000002. Okay, so I'm going to send Joe this not which my mailer hands off to my mail daemon. It looks to see if it has a "Joe Sixpack" key on its keyring. No? Put a request to the freenet for "Joe Sixpack" We receive the key and add to our keyring and encrypt our pending outgoing message with the public key held there in. We also request "Joe Sixpack-Seq" to get the current sequence number at the time of sending. If we insert and get a key collision we increment the sequence number and re-insert. if we don't then it is merrily upon its way. Meanwhile back at Joe's daemon, it gets a periodicly requested "Joe Sixpack-0000000000000000002". It increments "Joe Sixpack-Seq" and re-inserts it, decrypts the message and dumps it in Joes mailbox or whatever. So when you all get done laughing, please have pitty and realize I've been having a hell of a time figuring these fucking Keys out and set me straight. Kirk -- Kirk Reiser The Computer Braille Facility e-mail: ki...@br... University of Western Ontario phone: (519) 661-3061 |
From: Mark J. R. <mj...@st...> - 2001-02-22 02:24:14
|
On Wed, 21 Feb 2001, Kirk Reiser wrote: > Hi Folks: Well I was told that this is the place to be to discuss > mail specifications for the freenet. I have downloaded the current > cvs but there appears to be no documentation of how the current code Hey! The CVS WORKS! Yay for EOF! > works. Would someone mind giving a description of the protocal for > the current implementation? Dunno. > I will attempt to look through the code but have to admit I don't > really know java. I was hoping for something which was written in c. We could certainly implement the protocols in C. -- Mark Roberts mj...@st... |
From: Kirk R. <ki...@br...> - 2001-02-22 01:59:00
|
Hi Folks: Well I was told that this is the place to be to discuss mail specifications for the freenet. I have downloaded the current cvs but there appears to be no documentation of how the current code works. Would someone mind giving a description of the protocal for the current implementation? I will attempt to look through the code but have to admit I don't really know java. I was hoping for something which was written in c. Kirk |
From: Brandon <bl...@ut...> - 2001-02-12 20:53:12
|
> I have made some preliminary improvements to the website. A basic layout of > the new design is at eof.sourceforge.net/test, and a template file for it is > at eof.sourceforge.net/test/tmpl.shtml. > > Please tell me what you think. Neat. I like it. |
From: Timm M. <ad...@ma...> - 2001-02-12 18:52:13
|
I have made some preliminary improvements to the website. A basic layout of the new design is at eof.sourceforge.net/test, and a template file for it is at eof.sourceforge.net/test/tmpl.shtml. Please tell me what you think. |
From: Brandon <bl...@ut...> - 2001-02-12 06:41:03
|
I found the passwords. It should all be fixed now. |
From: Brandon <bl...@ut...> - 2001-02-12 06:17:14
|
> Mark J. Roberts told me that I should go and fix the EOF mailing lists > so that it includes a reply-to line in every post so that replies > automatically go to the mailing list instead of the person being > replied to. This was shown by how two of Brandon's posts went to me > directly rather than to the mailing list. Could someone with admin > for EOF fix this? I've tried to do it through ssh, but it isn't > obvious how to change mailing list configuration. Um...I don't seem to have the mailing lists passwords, actually. They're randomly generated upon mailing list creation. Luckily, they should be sent to me at the beginning of the month. As the ex-administrator of the Freenet mailing lists, I should know better. So I'll fix this when I get the passwords. |
From: Travis B. <be...@ex...> - 2001-02-12 04:34:57
|
Mark J. Roberts told me that I should go and fix the EOF mailing lists so that it includes a reply-to line in every post so that replies automatically go to the mailing list instead of the person being replied to. This was shown by how two of Brandon's posts went to me directly rather than to the mailing list. Could someone with admin for EOF fix this? I've tried to do it through ssh, but it isn't obvious how to change mailing list configuration. --=20 Yes, I know my enemies. They're the teachers who tell me to fight me. Compromise, conformity, assimilation, submission, ignorance, hypocrisy, brutality, the elite. All of which are American dreams. - Rage Against The Machine |
From: Brandon <bl...@ut...> - 2001-02-12 03:42:27
|
> I've been looking through the documentation for Java 1.1.8, and I can > find no obvious way to do string formatting. What I need this for is > converting date data into the date format used by the newest version > of fnnews, which requires padding numbers with zeros. This would be > very nontrivial only if Java had a good string formatting facility > like the standard C libraries or Python. So how do I do this without > having to code the whole thing by hand?! Java doesn't have a standard string formatting facility, but there are many freely available libraries that provide C printf-like functions. You can find them on freshmeat. Here's a link for one I stumbled upon earlier today: http://www.acme.com/java/software/Acme.Fmt.html Java has a lot of deficiencies, mainly in its standard libraries lacking obvious functions. String handling is the part which I've found most lacking. Luckily, there are freely available libraries. If you find some good functions, add them to eof.util.StringUtils. |
From: Travis B. <be...@ex...> - 2001-02-11 23:44:17
|
On Sun, Feb 11, 2001 at 04:38:00PM -0500, Travis Bemann wrote: > I've been looking through the documentation for Java 1.1.8, and I can > find no obvious way to do string formatting. What I need this for is > converting date data into the date format used by the newest version > of fnnews, which requires padding numbers with zeros. This would be > very nontrivial only if Java had a good string formatting facility > like the standard C libraries or Python. So how do I do this without > having to code the whole thing by hand?! Yes, I know that the answer is easy for simple stuff like this. However, I still want to know whether there is a real general purpose string formatting facility in Java. Padding integer strings is easy, but things are far more complicated with other formatting stuff. --=20 Yes, I know my enemies. They're the teachers who tell me to fight me. Compromise, conformity, assimilation, submission, ignorance, hypocrisy, brutality, the elite. All of which are American dreams. - Rage Against The Machine |
From: Travis B. <be...@ex...> - 2001-02-11 22:31:16
|
I've been looking through the documentation for Java 1.1.8, and I can find no obvious way to do string formatting. What I need this for is converting date data into the date format used by the newest version of fnnews, which requires padding numbers with zeros. This would be very nontrivial only if Java had a good string formatting facility like the standard C libraries or Python. So how do I do this without having to code the whole thing by hand?! --=20 Yes, I know my enemies. They're the teachers who tell me to fight me. Compromise, conformity, assimilation, submission, ignorance, hypocrisy, brutality, the elite. All of which are American dreams. - Rage Against The Machine |
From: Mark J. R. <mj...@st...> - 2001-02-11 21:50:21
|
On Sun, 11 Feb 2001, Travis Bemann wrote: > > Nevertheless, users don't want to sit there for a minutes waiting for the > > archive(s) and still-unarchived posts to come in. That's why IMHO you > > should cache recent posts, so the user can instantly access them. > > The archives are only created when up to or beyond a particular block > of posts exists. And note that you can still download all the new > posts after an archive. > > As for caching, that would not be a problem at all. This would make > it so that fnnews posts are downloaded automatically as one does > anything, and therefore make fnnews psychologically *appear* to be > faster. By using multithreading, fnnews can be made faster in > reality. By combining both of these things into a gateway, fnnews can > actually appear to not be sluggish. Well it wouldn't just be a psychological trick--caching posts in advance is simply a better approach, if you don't mind the little disk space and bandwidth it'd need. > > You should set the Reply-To field on this list IMHO. > > You mean that I should set my Reply-To field to the eof-dev list? No, you should configure the Eof-dev mailing list to set Reply-To: eo...@li... like the Freenet lists do. Basically the argument is that if you do, then people will accidentally post to the list when they want to send a private email, and if you don't, people will accidentally send private email when they want to post to the list. If you do, people will have to manually override the Reply-To field (most mailers support this with one keystroke), and if you don't, people will have to manually type in eo...@li... every time they send mail to the list. -- Mark Roberts mj...@st... |
From: Travis B. <be...@ex...> - 2001-02-11 21:39:56
|
On Sun, Feb 11, 2001 at 03:07:39PM -0500, Mark J. Roberts wrote: > On Sun, 11 Feb 2001, Travis Bemann wrote: >=20 > > On Sun, Feb 11, 2001 at 02:23:43PM -0500, Mark J. Roberts wrote: > > > On Sun, 11 Feb 2001, Travis Bemann wrote: > > > > > > > Here's my new hex-less version of fnnews, which is what I'm going to > > > > change the current implementation of freenet-news into (with > > > > archiving). > > > > By version of fnnews, I mean version of the standard of fnnews, not a > > version of software. > > > > > Are you planning to have the server cache new articles for configured > > > groups in advance? > > > > What server?! By server do you mean fnnews-NNTP gateway?! >=20 > Yes. NNTP client <-> fnnews server. Semantics. >=20 > > There is something similar to caching used by fnnews called archiving, > > which is essentially clients bundling fnnews ID blocks of size 64 into > > single files for quick downloading. >=20 > Nevertheless, users don't want to sit there for a minutes waiting for the > archive(s) and still-unarchived posts to come in. That's why IMHO you > should cache recent posts, so the user can instantly access them. The archives are only created when up to or beyond a particular block of posts exists. And note that you can still download all the new posts after an archive. As for caching, that would not be a problem at all. This would make it so that fnnews posts are downloaded automatically as one does anything, and therefore make fnnews psychologically *appear* to be faster. By using multithreading, fnnews can be made faster in reality. By combining both of these things into a gateway, fnnews can actually appear to not be sluggish. > > > On a completely unrelated note, debugging multithreaded applications = is > > > really hard. The backtraces are often meaningless and searches on Goo= gle > > > give me nothing useful... nevertheless, I have concurrent requests > > > semi-stable now, and inserts might have a bug or two in libfreenet. > > > > Depends. I've done some stuff with multithreading back in my Mac days > > and actually got everything to work fine. I've also done > > multithreading stuff with the ancient freenetlib in Python, but I had > > to cut that stuff out because it didn't work right. > > > > > I also have a really weird bug. My insert and request functions are > > > recursive--if a non-fatal error is encountered, they call themselves = and > > > decrement the retries argument (when it hits 0 it aborts). Anyway, the > > > request will always segfault when the recursion happens, even though = the > > > arguments are identical! WTF could this be? > > > > This is sort of the wrong context for this! Did you accidentally > > respond to the wrong post! >=20 > Nah, I'm tired and I'm hoping someone knows something somewhere. I've been > hacking pthreads for ages... >=20 > You should set the Reply-To field on this list IMHO. You mean that I should set my Reply-To field to the eof-dev list? --=20 Yes, I know my enemies. They're the teachers who tell me to fight me. Compromise, conformity, assimilation, submission, ignorance, hypocrisy, brutality, the elite. All of which are American dreams. - Rage Against The Machine |
From: Mark J. R. <mj...@st...> - 2001-02-11 20:21:23
|
On Sun, 11 Feb 2001, Travis Bemann wrote: > Here's my new hex-less version of fnnews, which is what I'm going to > change the current implementation of freenet-news into (with > archiving). Are you planning to have the server cache new articles for configured groups in advance? On a completely unrelated note, debugging multithreaded applications is really hard. The backtraces are often meaningless and searches on Google give me nothing useful... nevertheless, I have concurrent requests semi-stable now, and inserts might have a bug or two in libfreenet. I also have a really weird bug. My insert and request functions are recursive--if a non-fatal error is encountered, they call themselves and decrement the retries argument (when it hits 0 it aborts). Anyway, the request will always segfault when the recursion happens, even though the arguments are identical! WTF could this be? -- Mark Roberts mj...@st... |
From: Travis B. <be...@ex...> - 2001-02-11 20:04:28
|
Here's my new hex-less version of fnnews, which is what I'm going to change the current implementation of freenet-news into (with archiving). -- Yes, I know my enemies. They're the teachers who tell me to fight me. Compromise, conformity, assimilation, submission, ignorance, hypocrisy, brutality, the elite. All of which are American dreams. - Rage Against The Machine |
From: Brandon <bl...@ut...> - 2001-02-11 00:01:47
|
> How fnnews is implemented in the EOF CVS appears to be email/MIME > based. The thing is that fnnews isn't based on email and doesn't use > MIME. The reason why is that fnnews is quite different from email, > and fnnews doesn't use MIME because there is no reason to have images > and files in fnnews posts (they can just be stored in Freenet). I'm > thinking of completely reengineering the implementation of fnnews so > it isn't tied to email and MIME. Any comments? Tne news implementation doesn't rely on e-mail, it relies on the JavaMail library, which is a generic API for dealing with messages, folders, and the like. It certainly doesn't rely on MIME. I might have used MimeMessage instead of Message because it had a convenient utlity method that I found useful. However, it's not necessary. I think that switched from using a standard API like JavaMail to an completely non-standard API requires an explanation of why it is necessary and beneficial to do so. In other words, what do you gain by not using JavaMail? |
From: Travis B. <be...@ex...> - 2001-02-10 22:03:38
|
How fnnews is implemented in the EOF CVS appears to be email/MIME based. The thing is that fnnews isn't based on email and doesn't use MIME. The reason why is that fnnews is quite different from email, and fnnews doesn't use MIME because there is no reason to have images and files in fnnews posts (they can just be stored in Freenet). I'm thinking of completely reengineering the implementation of fnnews so it isn't tied to email and MIME. Any comments? --=20 Yes, I know my enemies. They're the teachers who tell me to fight me. Compromise, conformity, assimilation, submission, ignorance, hypocrisy, brutality, the elite. All of which are American dreams. - Rage Against The Machine |
From: Brandon <bl...@ut...> - 2001-02-10 20:28:50
|
> Anyway, I'd like to get it in the EOF CVS ASAP, so I want access. And > maybe if we're really really lucky those lazy GCJ guys will fix > BigInteger:divide so we can release real executables linked against > libfred.so, too! Cool. You now have access. A quick note on the CVS layout: the general plan is to stick your abstract/API stuff in one module and your protocol-specific stuff in another. So in your case you'd have something like a "versionControl" module and later a "cvs" module. The clients which you develop to use the abstract stuff separate from a specific protocol go in the first directory. So for instance for IRC I have a chat module which includes the little chat client I wrote and then an irc module. > Oh, and is anyone working on a bad-ass website layout? Do we want a > bad-ass website layout? I can write one, if you're interested. I'm fairly surely no one is working on the website. A bad-ass website layout would be very cool. There are intermittent permission problems with the website, so just make sure that you make any new files you create group read/writable. |
From: Mark J. R. <mj...@st...> - 2001-02-10 12:59:59
|
On Mon, 5 Feb 2001, Brandon wrote: > > Excellent. I'm going to start soon on my CVS-like (but not really) system > > for mapping a Freenet-published directory hierarchy into your filesystem > > so that you can download parts of it whenever you want. > > I think it would be way cool if you made actual CVS. :-) I'll probably be able to release this in a couple days. The only remaining issues are that metadata insertion is broken in libfreenet and I haven't written the parallel insert and request support. But otherwise: mjr::test$ chunga insert freenet:CHK@ somefiles/ I somefiles/anotherdir/onemore/xaa I somefiles/anotherdir/onemore/xab I somefiles/anotherdir/onemore/xac I somefiles/anotherdir/onemore/xad I somefiles/anotherdir/onemore/xae I somefiles/anotherdir/xaa I somefiles/anotherdir/xab I somefiles/anotherdir/xac I somefiles/anotherdir/xad I somefiles/anotherdir/xae I somefiles/anotherdir/xag I somefiles/xaf I somefiles/xag I somefiles/xah I somefiles/xai I somefiles/xaj I /tmp/chungatemp freenet:CHK@w5xOyzhw9-G0aPGSBMwWMYELbpgLAwE,V6jEKFzJ1eet9xXuiCbg0Q inserted. and then mjr::test$ rm -rf somefiles/ mjr::test$ chunga checkout freenet:CHK@w5xOyzhw9-G0aPGSBMwWMYELbpgLAwE,V6jEKFzJ1eet9xXuiCbg0Q G somefiles/anotherdir/onemore/xaa G somefiles/anotherdir/onemore/xab G somefiles/anotherdir/onemore/xac G somefiles/anotherdir/onemore/xad G somefiles/anotherdir/onemore/xae G somefiles/anotherdir/xaa G somefiles/anotherdir/xab G somefiles/anotherdir/xac G somefiles/anotherdir/xad G somefiles/anotherdir/xae G somefiles/anotherdir/xag G somefiles/xaf G somefiles/xag G somefiles/xah G somefiles/xai G somefiles/xaj and finally you download what you want (I'll request it all): mjr::test$ chunga get somefiles/ D somefiles/anotherdir/onemore/xaa D somefiles/anotherdir/onemore/xab D somefiles/anotherdir/onemore/xac D somefiles/anotherdir/onemore/xad D somefiles/anotherdir/onemore/xae D somefiles/anotherdir/xaa D somefiles/anotherdir/xab D somefiles/anotherdir/xac D somefiles/anotherdir/xad D somefiles/anotherdir/xae D somefiles/anotherdir/xag D somefiles/xaf D somefiles/xag D somefiles/xah D somefiles/xai D somefiles/xaj and they're there. I haven't started thinking about updating yet, but I don't think it'll be too hard. And of course once I have that done then we can start thinking about CVS. Anyway, I'd like to get it in the EOF CVS ASAP, so I want access. And maybe if we're really really lucky those lazy GCJ guys will fix BigInteger:divide so we can release real executables linked against libfred.so, too! Oh, and is anyone working on a bad-ass website layout? Do we want a bad-ass website layout? I can write one, if you're interested. -- Mark Roberts mj...@st... |