hvr%kimba@Sun.COM (Heather Rose) (09/01/88)
In article <1082@imagine.PAWL.RPI.EDU> jefu@pawl13.pawl.rpi.edu (Jeffrey Putnam) writes: >Me too. But I would like to add that I think that the window has >essentially closed. Alot of people said that UNIX would never catch on for alot of the same reasons that people are saying NeWS will not catch on: to unreliable, too hard to learn, too big. But, as the market has demanded more performance, UNIX has caught on more and more. I think the same thing could happen with NeWS. X11 has many very serious limitations that NeWS addresses. And as people need more functionality and do not find it in X[foo], then they will chose a new windowing system which could be NeWS or whatever else is the wizziest thing on the market. Besides, X is not the only other windowing system on the market today: SunView, Mac Windows, MS Windows, .... I think this group of people (USENET) is skewed towards X because of the type of people who read these groups. One thing X will have a big problem with is input from new devices (according to Jim Gettys at the Bay Area X User's Group meeting). NeWS has a very simple and clean input design which could be easily extended to receive input from any type of device. Another problem I see with X is lack of standards: no standard toolkit and no standard look and feel. Whenever we have a new rev of X, we also have to have a whole new slew of toolkits...mostly a complete re-write. NeWS will have more stability in the next release with NDE and Open Look. And even with future releases of NeWS...it will not change much because it has been relatively well-designed in the first place. Jim Gettys said that X was just a hacked together project that DEC latched on to. So even if a company were to write it's own toolkit in NeWS...that code would remain much more constant across releases than an X toolkit. One more point about NeWS: I think the biggest complaint aside from "too buggy" is that it is too difficult to understand for most programmers. I think that is the reason why we do not see 101 public domain NeWS toolkits on the market. Learning object-oriented programming and PostScript (a stack-based language) is a challenge. But this issue should be addressed by the new toolkit, NDE. We'll see what happens with windowing in the next year or so. Really, as far as the masses are concerned...they are still using glass tty's or little tiny Mac screens with Mac Windows. Most colleges still do not have bit-mapped displays for their undergrads (something other than Mac)...that is a vast market that is as yet untapped. I think it's premature to say that the window-wars are over and X has won. Geez... most colleges don't even offer a course on windowing systems... Most engineers still graduate only knowing FORTRAN as a programming language. Most people don't even read USENET or even know what it is... Personally, I think the window-wars will begin once machines like the Mac II become cheap enough for the masses to buy. And easy enough for the masses to use... Heather Rose disclaimer: of course these opinions are my own...nobody else would want them!
barnett@vdsvax.steinmetz.ge.com (Bruce G. Barnett) (09/01/88)
As a programmer, NeWS excites me more than SunView or X Windows. The flexibility allows for innovation in user interfaces. I have mentioned this before. NeWS is suffering from technology lag: Few are developing NeWS applications because there aren't many machines that have NeWS servers. Few companies are supporting NeWS servers because there aren't many applications. Catch-22. Until this is resolved, we won't see NeWS blossom. I can see several things that may change this: 1. NDE is so powerful that developing new applications will be trivial. 2. NeWS becomes a standard for the SysV.4 machines. Imagine if a 396-clone has a better window system than an Apollo/HP/DEC/etc workstation without NeWS? 3. There are several applications that demonstrate NeWS is more desirable. Examples might be: using NeWS over a 9600 baud modem. (I would LOVE to run my Sun window system at home on a $2000 computer with color, etc.) WYSIWYG - NeWS seems to be an ideal mechanism to interactively manipulate text and graphics. You can draw a diagram, and then paste it into a document. resizing - NeWS can redraw any window to any scale with little programming. Some applications might find this very desirable. 4. Hardware advances - if graphic accelerators can provide an abstraction at a higher level, drawing complex objects can be optimized more than simple pixels/lines. e.g. curved lines. But at this time, applications in X-windows are more portable, provided you are willing to maintain the software yourself. -- Bruce G. Barnett <barnett@ge-crd.ARPA> <barnett@steinmetz.UUCP> uunet!steinmetz!barnett
brt@homxc.UUCP (B.REYTBLAT) (09/01/88)
I'd like to address some of the misconceptions in the referenced article. In article <66437@sun.uucp>, hvr%kimba@Sun.COM (Heather Rose) writes: > In article <1082@imagine.PAWL.RPI.EDU> jefu@pawl13.pawl.rpi.edu (Jeffrey Putnam) writes: > >Me too. But I would like to add that I think that the window has > >essentially closed. > > Alot of people said that UNIX would never catch on for alot of the same > reasons that people are saying NeWS will not catch on: to unreliable, > too hard to learn, too big. But, as the > market has demanded more performance, UNIX has caught on more and more. > I think the same thing could happen with NeWS. X11 has many very serious > limitations that NeWS addresses. And as people need more functionality > and do not find it in X[foo], then they will chose a new windowing system > which could be NeWS or whatever else is the wizziest thing on the market. So far, so good. Here's where we part company: > Besides, X is not the only other windowing system on the market today: > SunView, Dead as a door nail. > Mac Windows, MS Windows, .... Not pertinent to the UNIX market. > I think this group of people > (USENET) is skewed towards X because of the type of people who read > these groups. May be. Maybe not. Data? > > One thing X will have a big problem with is input from new devices > (according to Jim Gettys at the Bay Area X User's Group meeting). NeWS > has a very simple and clean input design which could be easily extended > to receive input from any type of device. > > Another problem I see with X is lack of standards: no standard toolkit > and no standard look and feel. "One good thing about standards, is that there are so many to choose from": X Toolkit standards: MIT Xt Intrinsics approved in August MIT Core Widget Set spec being worked on even as we speak :-) X L&F standards: Athena (here & now) DECwindows (here & now) HP NewWave (here & now) OPEN LOOK (if and when it is actually available to the market) NeWS Toolkit standards: NDE (if and when it is actually available to the market) NeWS L&F standards: OPEN LOOK (if and when it is actually available to the market) Summary: X standards do exist in all but one area. NeWS standards have been delayed yet again (into next year now). >Whenever we have a new rev of X, we also > have to have a whole new slew of toolkits...mostly a complete re-write. That was true in X10.4 -> X11r1 and in X11r1 -> X11r2. It is NOT true for X11r2 -> X11r3. > NeWS will have more stability in the next release with NDE and Open Look. ^^^^ When NDE and OPEN LOOK are actually available, and on more than just Sun machines, come back we'll look at the problem again. > And even with future releases of NeWS...it will not change much because > it has been relatively well-designed in the first place. Jim Gettys said > that X was just a hacked together project that DEC latched on to. So even if a > company were to write it's own toolkit in NeWS...that code would remain > much more constant across releases than an X toolkit. Prove it. I can argue the opposite: - X has a large base of users and contributors concerned with portability. - NeWS has Joy, Gosling, and a few other people. Bright people, but few nonetheless. And they appear to be more concerned with selling iron than the MIT Consortium. > One more point about NeWS: I think the biggest complaint aside from > "too buggy" is that it is too difficult to understand for most programmers. > I think that is the reason why we do not see 101 public domain NeWS > toolkits on the market. Learning object-oriented programming and PostScript > (a stack-based language) is a challenge. But this issue should be > addressed by the new toolkit, NDE. In soc.women this argument is called "blaming the victim" :-) But seriously, folks, OO programming is actually easier to teach then other forms. I've done it. [ more stuff deleted ] > > Heather Rose > > disclaimer: of course these opinions are my own...nobody else would > want them! Editorial : - I actually think NeWS is a better way of doing things than X - The dual X11/NeWS server is the best approach to combining X with another window system. - I like Sun HW. I run a network of several machines for myself and others. I've recommended them to others. - I think Sun is digging its own grave by the following actions: + Denigrading X (as in the article referenced) + Refusing to deliver a product quality X11 server to the market in a timely way. I am aware of the promised delivery of the X11/NeWS server. Its not soon enough. I need a fast server yesterday. + Refusing to deliver an X based OPEN LOOK implementation, source compatible with other implementations of OPEN LOOK. These actions might have gone by without consequences, had there not been viable alternatives to consider: + Sony workstations + DEC workstations + Tektronix workstations + HP workstations + Compaqs with a 3rd party color board and awsomely fast servers. this list is in no particular order. - A better way of introducing NeWS would have been: + Become a leader in the X market by delivering the best price/performance X11 server (Sun is quite capable of doing it) + Introduce the dual server about 1.5-2 years later as added functionality. Ben Reytblat homxc!brt Disclaimer: These opinions are my own. They do not represent anyone's views but my own. I'm not connected with any organization making policy decisions.
jg@jumbo.dec.com (Jim Gettys) (09/01/88)
In article <66437@sun.uucp> hvr@sun.UUCP (Heather Rose) writes: >In article <1082@imagine.PAWL.RPI.EDU> jefu@pawl13.pawl.rpi.edu (Jeffrey Putnam) writes: >>Me too. But I would like to add that I think that the window has >>essentially closed. > >Alot of people said that UNIX would never catch on for alot of the same >reasons that people are saying NeWS will not catch on: to unreliable, >too hard to learn, too big. But, as the >market has demanded more performance, UNIX has caught on more and more. >I think the same thing could happen with NeWS. X11 has many very serious >limitations that NeWS addresses. And as people need more functionality >and do not find it in X[foo], then they will chose a new windowing system >which could be NeWS or whatever else is the wizziest thing on the market. NeWS has serious limitations itself. I used to be paid to write code that X supports easily, and NeWS is not able to support at all. I personally consider X's limitations, such as they are, less of a problem than NeWS limitations, as they are. > >One thing X will have a big problem with is input from new devices >(according to Jim Gettys at the Bay Area X User's Group meeting). NeWS >has a very simple and clean input design which could be easily extended >to receive input from any type of device. What I have said on numerous occasions is that living with broken "features" is a bad idea. In the early V11 design, we included support for multiple input devices, but found when we implemented the design that it was seriously flawed. Rather than live with a bad design, we removed this support in the core X specification. People are working on X extensions now to "get it right", and support your favorite knobs, dials, buzz boxes, etc. For example, look at the Digital VS8000, which has quite a complex button box. I would hope that this becomes standardized soon; there are proposals on the table for this now. You are invited to review them. > >Another problem I see with X is lack of standards: no standard toolkit >and no standard look and feel. Whenever we have a new rev of X, we also >have to have a whole new slew of toolkits...mostly a complete re-write. >NeWS will have more stability in the next release with NDE and Open Look. The toolkit intrinsics are now quite stable and available. As for look and feel, there are competing "standards"; only time will tell which of Xui, Open Look, etc. become the true standard. But standards form when people agree, and declaring something a "standard" does not make it "the standard" until such agreement is reached. >And even with future releases of NeWS...it will not change much because >it has been relatively well-designed in the first place. Jim Gettys said >that X was just a hacked together project that DEC latched on to. So even if a >company were to write it's own toolkit in NeWS...that code would remain >much more constant across releases than an X toolkit. I've never said X was a hacked together project DEC latched onto; in no location, at no time. X up through version 10 was what a very small number (approximately 2-3) people could do. It was successful beyond our wildest dreams. But rather than allowing something that was seriously limited become so widely used that it could not be dislodged, we completley redesigned X in version 11, and completely reimplemented it. The redesign was done completely openly, with input from many talented people in many companies. This is hardly "hacked up". > >Heather Rose > >disclaimer: of course these opinions are my own...nobody else would > want them! Yup. And I'd recommend listening to what people actually say before you quote someone. Jim Gettys Digital Equipment Corporation Systems Research Center
brian@hcx1.SSD.HARRIS.COM (09/02/88)
> One thing X will have a big problem with is input from new devices There hasn't been a flood of new devices over the past ten years, with the exception of "specialty" devices (eg tablets for CAD'ers). Basically, you've got the keyboard, and you've got the mouse. Until we grow more hands or evolve differently, we probably won't see many new input devices. True, NeWS has the advantage here, but I don't see it tipping the scales much. > Another problem I see with X is lack of standards: no standard toolkit > and no standard look and feel. Whenever we have a new rev of X, we also > have to have a whole new slew of toolkits...mostly a complete re-write. > NeWS will have more stability in the next release with NDE and Open Look. Open Look won't be available over X? It really depends on which user interface becomes the standard. > We'll see what happens with windowing in the next year or so. Really, > as far as the masses are concerned...they are still using glass tty's A good point - I wonder how long it will take before we're rid of them? The winner of the Windows War will not be decided by technical merit (I'd give it to NeWS there) or ease of use -- it will be decided by the vendors writing the applications. The window system should be fairly transparent - it's the application the customer or programmer really cares about. If I were so bold to predict a winner, I'd pick X. ----- brian email: brian@hcx1.SSD.HARRIS.COM
hvr%kimba@Sun.COM (Heather Rose) (09/02/88)
In article <13317@jumbo.dec.com> jg@jumbo.UUCP (Jim Gettys) writes: > >I've never said X was a hacked together project DEC latched onto; in >no location, at no time. You are of course right you never said those words. What I interpreted as "hacked up" was the X10 project that 2-3 people worked on with much success. X11 was a complete re-design as far as I know that DEC wanted very knowledgeable people like yourself to fully re-think through and implement. What I see as "lost effort" so to speak are the applications written for X10. I understand HP has a very nice implementation of X10 with many goodies. I don't want to seem "anti-X11" since it is something I will be using alot of in the near future. I have used X10 and liked it quite a bit. I was very impressed with that work. >Yup. And I'd recommend listening to what people actually say before >you quote someone. If I offended you, I am very sorry. It was never my intention to do so. Thanks for taking the time to educate me, I appreciate it! Heather Rose disclaimer: these opinions are mine, mine, all mine!
roy@phri.UUCP (Roy Smith) (09/03/88)
brian@hcx1.SSD.HARRIS.COM writes: > There hasn't been a flood of new devices over the past ten years, with > the exception of "specialty" devices (eg tablets for CAD'ers). Basically, > you've got the keyboard, and you've got the mouse. Until we grow more > hands or evolve differently, we probably won't see many new input devices. This is probably circular reasoning. You claim we don't have to support multiple input devices because they don't exist. I claim they don't exist because they are hard to support. In addition to mice and tablets (essentially the same thing), people like Silicon Graphics already support dial-and-button boxes as standard I/O (well, mostly I) devices. Touch screens are becomming popular; I use one every time I use a cash machine. And, if you want esoterica, try things like Richard Feldman's joystring, a sort of 6-degree-of-freedom-with-tactile-feedback 3-d joystick (designed for use in pairs, one per hand). -- Roy Smith, System Administrator Public Health Research Institute {allegra,philabs,cmcl2,rutgers}!phri!roy -or- phri!roy@uunet.uu.net "The connector is the network"
spock@ihlpf.ATT.COM (Weiss) (09/06/88)
In article <13317@jumbo.dec.com> jg@jumbo.UUCP (Jim Gettys) writes: >NeWS has serious limitations itself. I used to be paid to write code that >X supports easily, and NeWS is not able to support at all. I personally >consider X's limitations, such as they are, less of a problem than >NeWS limitations, as they are. I have seen a number of articles that give X's shortcomings, but I can't think of one that gives NeWS's. Can you give the NeWS limitations? -- Ed Weiss "I thought it was generally accepted, sir, that att!ihlpf!spock vulcans are an advanced and most honorable race." "They are, they are. And damn annoying at times."
davidh@dent.Berkeley.EDU (David S. Harrison) (09/07/88)
> I have seen a number of articles that give X's shortcomings, but > I can't think of one that gives NeWS's. Can you give the NeWS > limitations? 1. No primitives for modifying colormaps. 2. No support for plane masks on multi-plane devices. 3. No support for interlocked tiling of filled regions. David Harrison UC Berkeley Electronics Research Lab (davidh@ic.Berkeley.EDU) (...!ucbvax!ucbcad!davidh)
barnett@vdsvax.steinmetz.ge.com (Bruce G. Barnett) (09/07/88)
In article <6002@ihlpf.ATT.COM> spock@ihlpf.UUCP (Ed-Weiss) writes: |Can you give the NeWS limitations? When displaying VLSI layers, it is convenient to have translucent colors, so you can see all of the layers, and not just the top. I realize that rasterops on colored areas don't really make sense, but there are some packages that make use of this "feature". (I agree. This does seem a hack.) What is the best way to use NeWS to accomplish this? -- Bruce G. Barnett <barnett@ge-crd.ARPA> <barnett@steinmetz.UUCP> uunet!steinmetz!barnett
roy@phri.UUCP (Roy Smith) (09/07/88)
In article <6002@ihlpf.ATT.COM> spock@ihlpf.UUCP (Ed-Weiss) writes: > I have seen a number of articles that give X's shortcomings, but > I can't think of one that gives NeWS's. Can you give the NeWS limitations? How about, "it's slow". I should qualify that; NeWS 1.1 is slow on a 4-Mbyte 3/50 because it pages so much. It works just dandy on a 8-Mbyte 3/160. Whether this bloatatiousness is inherent in NeWS or just an implementation problem remains to be seen. My guess is that it should be possible to implement NeWS in a much smaller package, but that's just a guess. Sun may be good at some things, but writing small programs isn't one of them. -- Roy Smith, System Administrator Public Health Research Institute {allegra,philabs,cmcl2,rutgers}!phri!roy -or- phri!roy@uunet.uu.net "The connector is the network"