On Wed, 26 Jul 2000, you wrote: <> > > While I still don't believe that discussion forums will work (see my earlier > > post which you did not respond to), I don't think that specialized clients > > for > > everysingle ruleset sounds good at all. > > I can't find it - could you sent it to me offlist?
re: RoadMap 18:08:53 (GMT +7:00) Tuesday, Jul 25 (that would be 12:08:53 for you if you are still in London, 4:08 or 5:08 if you are in California). > > If the current "everything on the web" thing (started by Hotmail) tells us > > anything it is that people do not like to download specialized clients. > > Oh come on, there is no basis for that claim. Hotmail is popular > because it just-so-happens that reading and sending email over the web > works reasonably well. I could point to the success of Napster, Winamp, > Netscape mail & news, and a whole host of clients that people do > download and do use. AFAIK, Winamp comes with AOL. Netscape has a big issue with that people don't want to download it (that was what that whole hoopla about Microsoft was about). Napster may be an exception, but it is also the only such exception since ICQ was new. > > Anyways, by general public I mean the general public that I give a damn > > about > > (ie the Linux/geek computer literrate general public who can write their own > > shell scripts). > > The language is pretty easy to use, personally I am very comfortable in > coding in RPN directly (as are any people - just look at how many people > use RPN calculators), but it would be trivial to write a parser to > convert between RPN and Polish Notation, or a Lisp-style layout. Better > still, a graphical tool could be created relatively easily which would > allow the design of Sesil subspace definitions. So why not take the full step and make it for machine read only? <> > > I meant, do you want a human readable code or easily machine readable. In > > other > > words, is check signature CHECK_SIGNATURE() or 0x4A? > > Ah, yes, it is all human-readable (as you will see if you take a look at > Freenet/subspaces/Sesil.java). Since it will be read by nodes more then people, I think it should be machine readable. > > If you want to use a language that most users won't want to write for > > manually > > anyways, I would suggest the latter as we want to make things as easy for > > the > > nodes as possible, and not waste bandwidth. > > Since only inserts will need to contain a copy of the language, I don't > really think bandwidth will be much of an issue, most Sesil programs > will probably be less than 1k in length, when compaired to 6 or 7 MB > files which will probably be common on Freenet, I don't think there is > any need to worry too much about this. Using a human-readable syntax > also makes the language much easier to extend and doesn't place any 256 > command limits on things. DataInserts and DataReply-s would need to contain the language. And I don't think we will have 6-7 MB files, since we have said we would (and should if we have any form of efficiency in mind) split files into much smaller chunks then that. Also, the script would be in an SVK like document - ie, containing only a reference to the actual data - and it is important to performance to keep those small. A byte code does not have to have a 256 command limit, one could easily use an encoding like UTF-8 to allow an arbitrary number of functions. > If you are really that paranoid about bandwidth then forget Sesil, there > are many ways that we could redesign the message format to make it more > compact ;-) Node performance is of bigger concern. > More importantly, I have already implemented the human-readable parser > ;-) I have thrown out thousands of lines of code that I have written for Freenet because we changed our minds. > Ian. > > _______________________________________________ > Freenet-dev mailing list > Freenet-dev at lists.sourceforge.net > http://lists.sourceforge.net/mailman/listinfo/freenet-dev -- \oskar _______________________________________________ Freenet-dev mailing list Freenet-dev at lists.sourceforge.net http://lists.sourceforge.net/mailman/listinfo/freenet-dev
