|
From: Josh S. <jo...@ed...> - 2002-04-12 19:15:36
|
http://makeashorterlink.com/ so, i still think about the u9k a lot... but am still undecided if it is worth the effort as there are many other similar projects. i think the only thing that differentiates the u9k from the others is the statistical analysis ideas, which are pretty cool, but i would like the u9k to coexist with other similar projects as much as possible without reimplementing their work. Bookie is a remarkably similar project, though I think that it is as about as well maintained as the u9k... basically, is the u9k worht a rewrite? i'm not sure... ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Steven M. <sc...@ed...> - 2002-04-12 19:37:07
|
I've been thinking about it lately too. One idea I had was to strip the server down to simple servlets that maintain a hieriarchy (using the file system) of RDF files. Then writing a mozilla client that uses the RDF via simple HTTP. Using the RDF handling in Mozilla, we could combine serveral files (an RDF datasource can be multiple files) and basically do database operations on them (filtering and sorting). -s Josh Steiner wrote: > http://makeashorterlink.com/ > > so, i still think about the u9k a lot... but am still undecided if it is > worth the effort as there are many other similar projects. i think the > only thing that differentiates the u9k from the others is the statistical > analysis ideas, which are pretty cool, but i would like the u9k to coexist > with other similar projects as much as possible without reimplementing > their work. Bookie is a remarkably similar project, though I think that > it is as about as well maintained as the u9k... basically, is the u9k > worht a rewrite? i'm not sure... > > ________________________________________________________________ > Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi > > > _______________________________________________ > Ultra9000-devel mailing list > Ult...@li... > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel |
|
From: Josh S. <jo...@ed...> - 2002-04-12 20:13:54
|
> RDF via simple HTTP. Using the RDF handling in Mozilla, we could combine > serveral files (an RDF datasource can be multiple files) and basically > do database operations on them (filtering and sorting). so you are saying do the sorting and filtering in hte client? > > -s > > Josh Steiner wrote: > > http://makeashorterlink.com/ > > > > so, i still think about the u9k a lot... but am still undecided if it is > > worth the effort as there are many other similar projects. i think the > > only thing that differentiates the u9k from the others is the statistical > > analysis ideas, which are pretty cool, but i would like the u9k to coexist > > with other similar projects as much as possible without reimplementing > > their work. Bookie is a remarkably similar project, though I think that > > it is as about as well maintained as the u9k... basically, is the u9k > > worht a rewrite? i'm not sure... > > > > ________________________________________________________________ > > Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi > > > > > > _______________________________________________ > > Ultra9000-devel mailing list > > Ult...@li... > > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel > > > > _______________________________________________ > Ultra9000-devel mailing list > Ult...@li... > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel > ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Steven M. <sc...@ed...> - 2002-04-12 20:55:36
|
Josh Steiner wrote: >>RDF via simple HTTP. Using the RDF handling in Mozilla, we could combine >>serveral files (an RDF datasource can be multiple files) and basically >>do database operations on them (filtering and sorting). > > > so you are saying do the sorting and filtering in hte client? > Basically. Some of the filtering would be done by choosing which RDF files to request, and some would be done with Mork in Mozilla (Mork is the interal database like thing used for mail, news, and other RDF datasources.) -s |
|
From: Josh S. <jo...@ed...> - 2002-04-12 21:15:38
|
I'm not too fond of puting this logic into the client itself, multiple front ends (ie, an html one like we have today) is really a key feature. the way I envision this is that our backend server can output the nessesary RDF for mozilla to intergreate into the existing mozilla bookmark UI, i think this is what bookie does, or aims to do... i'm super busy right now, so excuse any incoherence on my part... On Fri, 12 Apr 2002, Steven McDonald wrote: > > > Josh Steiner wrote: > >>RDF via simple HTTP. Using the RDF handling in Mozilla, we could combine > >>serveral files (an RDF datasource can be multiple files) and basically > >>do database operations on them (filtering and sorting). > > > > > > so you are saying do the sorting and filtering in hte client? > > > > Basically. Some of the filtering would be done by choosing which RDF > files to request, and some would be done with Mork in Mozilla (Mork is > the interal database like thing used for mail, news, and other RDF > datasources.) > > -s > > > _______________________________________________ > Ultra9000-devel mailing list > Ult...@li... > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel > ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Steven M. <sc...@ed...> - 2002-04-12 21:38:03
|
Josh Steiner wrote: > I'm not too fond of puting this logic into the client itself, multiple > front ends (ie, an html one like we have today) is really a key feature. The hard work is already done on the client. In fact, adding complexity on the server doesn't really make the client any simpler, since will be using the RDF stuff in mozilla anyway (filtering an RDF datasource is like 3 lines of XUL). > the way I envision this is that our backend server can output the > nessesary RDF for mozilla to intergreate into the existing mozilla > bookmark UI, i think this is what bookie does, or aims to do... > There's no reason we couldn't impliment this idea so that it integrates with the existing ui. > i'm super busy right now, so excuse any incoherence on my part... > To be clear, I'm not saying we *should* do this, it's just an idea. -s |
|
From: Jonathan N. <ph...@ed...> - 2002-04-12 21:43:34
|
as a user, :-) i would just like to restate my wanting some kind of categorization feature. my u9k has reached the point that it's difficult for me to find what i'm looking for if it's neither popular nor unpopular/new... so i use it less now. anyway, i'll shut up now. -j |
|
From: Josh S. <jo...@ed...> - 2002-04-12 21:47:10
|
the point of this thread is that there are other things out there that will do simple remote server bookmarking and do it much better than the u9k, i dont see the point of spending dozens of hours continuing to work on something that is so pointless just because _I_ wrote it and mine has a better name... there must be another reason to compell me onto 0.2 ... i think that reason has to be "thre is a place for the u9k that fits into the web of other similar projects" ... the niche exists, we just need to define it carefully On Fri, 12 Apr 2002, Jonathan Nelson wrote: > > > as a user, :-) > > i would just like to restate my wanting some kind of categorization > feature. my u9k has reached the point that it's difficult for me to find > what i'm looking for if it's neither popular nor unpopular/new... so i use > it less now. > > anyway, i'll shut up now. > > -j > > > _______________________________________________ > Ultra9000-devel mailing list > Ult...@li... > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel > ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Jonathan N. <ph...@ed...> - 2002-04-12 21:55:03
|
ok cool, that makes sense. do you know how many hits/day u9k gets? On Fri, 12 Apr 2002, Josh Steiner wrote: > the point of this thread is that there are other things out there that > will do simple remote server bookmarking and do it much better than the > u9k, i dont see the point of spending dozens of hours continuing to work > on something that is so pointless just because _I_ wrote it and mine has a > better name... there must be another reason to compell me onto 0.2 ... i > think that reason has to be "thre is a place for the u9k that fits into > the web of other similar projects" ... the niche exists, we just need to > define it carefully > > On Fri, 12 Apr 2002, Jonathan Nelson wrote: > > > > > > > as a user, :-) > > > > i would just like to restate my wanting some kind of categorization > > feature. my u9k has reached the point that it's difficult for me to find > > what i'm looking for if it's neither popular nor unpopular/new... so i use > > it less now. > > > > anyway, i'll shut up now. > > > > -j > > > > > > _______________________________________________ > > Ultra9000-devel mailing list > > Ult...@li... > > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel > > > > ________________________________________________________________ > Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi > |
|
From: Josh S. <jo...@ed...> - 2002-04-12 22:06:19
|
i dont know On Fri, 12 Apr 2002, Jonathan Nelson wrote: > > ok cool, that makes sense. > > do you know how many hits/day u9k gets? > > > > On Fri, 12 Apr 2002, Josh Steiner wrote: > > > the point of this thread is that there are other things out there that > > will do simple remote server bookmarking and do it much better than the > > u9k, i dont see the point of spending dozens of hours continuing to work > > on something that is so pointless just because _I_ wrote it and mine has a > > better name... there must be another reason to compell me onto 0.2 ... i > > think that reason has to be "thre is a place for the u9k that fits into > > the web of other similar projects" ... the niche exists, we just need to > > define it carefully > > > > On Fri, 12 Apr 2002, Jonathan Nelson wrote: > > > > > > > > > > > as a user, :-) > > > > > > i would just like to restate my wanting some kind of categorization > > > feature. my u9k has reached the point that it's difficult for me to find > > > what i'm looking for if it's neither popular nor unpopular/new... so i use > > > it less now. > > > > > > anyway, i'll shut up now. > > > > > > -j > > > > > > > > > _______________________________________________ > > > Ultra9000-devel mailing list > > > Ult...@li... > > > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel > > > > > > > ________________________________________________________________ > > Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi > > > > ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Jonathan N. <ph...@ed...> - 2002-04-12 22:10:07
|
> do you know how many hits/day u9k gets? the reason for this question is probably obvious - i'm sitting here drooling in front of the google API. the backlink and relation functions could fit well into a u9k derivative. [link:, related:]. etc. it occurs to me that it would be possible to do diffs across their results with and without safesearch enabled. .... heeheehee... -j |
|
From: Josh S. <jo...@ed...> - 2002-04-12 22:15:28
|
> the reason for this question is probably obvious - i'm sitting here > drooling in front of the google API. > > the backlink and relation functions could fit well into a u9k derivative. > [link:, related:]. how is this useful to u9k? > it occurs to me that it would be possible to do diffs across their results > with and without safesearch enabled. .... heeheehee... what is safesearch? ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Steven M. <sc...@ed...> - 2002-04-12 23:01:16
|
Josh Steiner wrote: >>the reason for this question is probably obvious - i'm sitting here >>drooling in front of the google API. >> >>the backlink and relation functions could fit well into a u9k derivative. >>[link:, related:]. > > > how is this useful to u9k? We could easily provide google's list of sites that link to a given page, or google's list of pages related to a give page, though I think we could give the same thing by sending them to google with the right URL. > > >>it occurs to me that it would be possible to do diffs across their results >>with and without safesearch enabled. .... heeheehee... > > > what is safesearch? > The system that filters porn out of your search results. -s |
|
From: Josh S. <jo...@ed...> - 2002-04-12 23:15:46
|
> > what is safesearch? > > > > The system that filters porn out of your search results. hmm... then i need unsafesearch so i get MORE porn in my results! ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Jonathan N. <ph...@ed...> - 2002-04-12 23:17:52
|
that's exactly what the diff scheme would give you!!! easy, too... :-p On Fri, 12 Apr 2002, Josh Steiner wrote: > > > what is safesearch? > > > > > > > The system that filters porn out of your search results. > > hmm... then i need unsafesearch so i get MORE porn in my results! > > ________________________________________________________________ > Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi > > > _______________________________________________ > Ultra9000-devel mailing list > Ult...@li... > https://lists.sourceforge.net/lists/listinfo/ultra9000-devel > |
|
From: Jonathan N. <ph...@ed...> - 2002-04-12 23:13:52
|
> > the backlink and relation functions could fit well into a u9k derivative. > > [link:, related:]. > > how is this useful to u9k? you could use google's referential scheme to present related stuff. so if someone accumulates a few sites that are related, you can suggest other things that are related by subject but not explicit links. "you're obviously looking for this sort of thing, but you won't find links to these other sites in the ones you've got already." the subject-relational part is pretty significant in terms of effort, but might be possible to search through for keywords, then search the keywords via google, and check correpondences between result sets. yeah, i'm sure that's way too heavy and inelegant for a lot of different reasons. :-) i have no idea where you're going with this, but it's an idea. > > it occurs to me that it would be possible to do diffs across their results > > with and without safesearch enabled. .... heeheehee... > > what is safesearch? adult content filter. this has nothing to do with u9k, but you could generate results that only list adult content without too much trouble. -j |
|
From: Josh S. <jo...@ed...> - 2002-04-13 00:36:58
|
> > I'm not too fond of puting this logic into the client itself, multiple > > front ends (ie, an html one like we have today) is really a key feature. > > The hard work is already done on the client. In fact, adding complexity > on the server doesn't really make the client any simpler, since will be > using the RDF stuff in mozilla anyway (filtering an RDF datasource is > like 3 lines of XUL). Then maybe I'm not following you, what sort of filtering would you be doing in XUL? is this just choosing which sets of RDF to display? if you are talking about building the lists/boxes/folders this definitely needs to be server side if we are going to have multiple frontends. > > the way I envision this is that our backend server can output the > > nessesary RDF for mozilla to intergreate into the existing mozilla > > bookmark UI, i think this is what bookie does, or aims to do... > > > > There's no reason we couldn't impliment this idea so that it integrates > with the existing ui. that would be ideal. > > i'm super busy right now, so excuse any incoherence on my part... > > > > To be clear, I'm not saying we *should* do this, it's just an idea. yeah, I'm just trying to understand you because i dont think i get your idea fully ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Steven M. <sc...@ed...> - 2002-04-13 00:49:06
|
Josh Steiner wrote: >>>I'm not too fond of puting this logic into the client itself, multiple >>>front ends (ie, an html one like we have today) is really a key feature. >> >>The hard work is already done on the client. In fact, adding complexity >>on the server doesn't really make the client any simpler, since will be >>using the RDF stuff in mozilla anyway (filtering an RDF datasource is >>like 3 lines of XUL). > > > Then maybe I'm not following you, what sort of filtering would you be > doing in XUL? Advanced stuff like "all links to mozilla.org" or something like sorting by date or number of visits. > is this just choosing which sets of RDF to display? if you > are talking about building the lists/boxes/folders No, this would be done by having different folders (or RDF files) for each user and user's folders (for example there might be an scm directory that had an file ultra.rdf which would be my top level bookmarks, with different folders stored below that). If you wanted to search over all your bookmarks then you would add all your user RDF files to one datasource and use mozilla to filter (i.e. search) that. Or you could pull in all users and search over that. You definately wouldn't be pulling the whole database down each time. > this definitely needs > to be server side if we are going to have multiple frontends. > I thought one of your ideas was to move to a mozilla only frontend. I think that was the last one you shared with me at least :-) -s |
|
From: Josh S. <jo...@ed...> - 2002-04-13 00:55:33
|
> I thought one of your ideas was to move to a mozilla only frontend. I > think that was the last one you shared with me at least :-) yay miscommunication, i dont think i ever wanted only moz frontend, but lets leave history, i think its important to have multiple front ends, for instance if i sit down at some random web cafe computer i wont be able to install moz but i want ot have access to my bookmarks. ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Steven M. <sc...@ed...> - 2002-04-13 01:18:01
|
Josh Steiner wrote: >>I thought one of your ideas was to move to a mozilla only frontend. I >>think that was the last one you shared with me at least :-) > > > yay miscommunication, i dont think i ever wanted only moz frontend, but > lets leave history, i think its important to have multiple front ends, for > instance if i sit down at some random web cafe computer i wont be able to > install moz but i want ot have access to my bookmarks. > This idea doesn't make other front ends impossible, but maybe a little more complicated. I believe that there are Java classes (or beans) for dealing with RDF. There are probably perl modules too. If not, it's not a terribly complicated format. Anyway, this idea came about one day when I was trying to think of how one would impliment ultra to be scalable, have folders, and not use a database on the server side (without basically writing a database). A simple web fronted could just XSLT the RDF files in to XHTML. The advantage to this idea is that the (at least mozilla) client would work from static pages, and servlets would only be needed to add or remove links. This would also hold true for a Java Applet or native client. -s |
|
From: Josh S. <jo...@ed...> - 2002-04-13 01:34:19
|
> This idea doesn't make other front ends impossible, but maybe a little > more complicated. I believe that there are Java classes (or beans) for > dealing with RDF. There are probably perl modules too. If not, it's not > a terribly complicated format. i'm still trying to follow you here, but honestly i think having a db backend is pretty desirable. i would like to do statistical analysis on hitcount vs time, so if we were putting a timestamp for every visit into, say, and rdf stream to get filtered at the XUL level our data stream will bloat quick. i'm tending to think that having folder just be some meta data wrapped around an sql query string is graceful, easy and very powerful. a servlet would slap some appropriate frontend formatting to the query result (html, xul, libglade) and thats about it. > Anyway, this idea came about one day when I was trying to think of how > one would impliment ultra to be scalable, have folders, and not use a > database on the server side (without basically writing a database). your idea seems pretty cool, but i think its ok to have a db backend for this. if you can setup apache/tomcat you can setup mysql. ________________________________________________________________ Josh .. Yoshi .. Joschi .. xiphoidprocess.com .. eds.org/~joschi |
|
From: Steven M. <sc...@ed...> - 2002-04-13 02:32:19
|
Josh Steiner wrote: >>This idea doesn't make other front ends impossible, but maybe a little >>more complicated. I believe that there are Java classes (or beans) for >>dealing with RDF. There are probably perl modules too. If not, it's not >>a terribly complicated format. > > > i'm still trying to follow you here, but honestly i think having a db > backend is pretty desirable. i would like to do statistical analysis on > hitcount vs time, so if we were putting a timestamp for every visit into, > say, and rdf stream to get filtered at the XUL level our data stream will > bloat quick. You could easily store that info in a seperate RDF (just like you'd probably store it in a different table in a Database) and only access it when needed. Of course, that data is already stored on the system in the Apache logs too. (to answer Jon's question, looking at Apr 7 - Apr 11, Ultra gets about 120 hits/day; high 155, low 85. not much data, I know) > > i'm tending to think that having folder just be some meta data wrapped > around an sql query string is graceful, easy and very powerful. a servlet > would slap some appropriate frontend formatting to the query result (html, > xul, libglade) and thats about it. > I was actually thinking (more along what's be "conventional" for ultra) that everything would be in the database, including the hierarchy and any meta data. > >>Anyway, this idea came about one day when I was trying to think of how >>one would impliment ultra to be scalable, have folders, and not use a >>database on the server side (without basically writing a database). > > > your idea seems pretty cool, but i think its ok to have a db backend for > this. if you can setup apache/tomcat you can setup mysql. > Yea, it was more of a thought experiment than anything else. There's nothing I can think of in Ultra that generates enough data to warrent a real database (other than hit tracking, which I wasn't thinking of at all at the time). I can't imagine coping with a folder with more than 2000 links in it... On the other hand, using something database like that we could toss the data in to would make programming easier, and wouldn't have the overhead of a "real" database (That's pretty much what MySQL is). -s |