Follow Slashdot blog updates by subscribing to our blog RSS feed

 



Forgot your password?
typodupeerror
×
X GUI

XFree86 Release Plans 152

sfid writes "Just read at XFree86 about the release plans for 4.0. The first beta will be availiable in July, further on there will be releases every 4-6 weeks. " Mentions several new chipsets in the 3.3.x tree, as well as several interesting new features for the 4.0 tree including video in a window, multihead, integrated TrueType, as well as 3D support Precision Insight's DRI stuff (which looked really excellent at LinuxExpo), Mesa, or SGIs GLX.
This discussion has been archived. No new comments can be posted.

XFree86 Release Plans

Comments Filter:
  • Hmm.. Jurassic Park?

    "It's a Unix system! I know this!"

  • I for one am cheering for such a major upgrade. With recent advances with Redhat 6.0, GNOME 1.0, GTK+ 1.2, and many others (including the anticipated release of Mozilla), Linux is finally advancing far enough and fast enough to be a serious contender in the desktop market.

    It's important to note the large availiability of applications and tools not only to make it easy for developers to create products for Linux, but also the tools to make it easy for "normal" users to experience the advantages of using an open source OS.

    Especially with all the recent news coverage that Linux has been getting, the idea that device support (at least video device support) has started to become largely comprehensive (especially with a section of the market Linux has long been bad with, totally new drivers, as evidences by nVidia and Creative's moves) really adds to the appeal of the operating system.

    Quite simply, I cannot wait until I can get my hands on a version of XFree 4.0, especially if there's some very cool and useful features such as multi-head support, support for more cards, &c.

    Thanks to everyone that has helped develop and test X, and remember to support open source software!
  • I can run XFree86 on Windows NT using the Interix Posix implementation.

    You mean you can actually run one of the servers, or just that you can use the client libraries? Do the XFree86 client libraries differ from the stock X consortium ones in any significant way?
  • Hopefully they will fix some of the obvious memory leaks in XFree86. Like the XWrapper eatin my memory like I eat fried chicken.
  • Quake and QII have always had hooks for all sorts of mods. And Q3 will be cross platform.

    Why couldn't one just use this environment and add basic Desktop functionality? File browser as real objects... Web browser... Some sort of xterm...

    Or maybe it is easier to start from scratch. Give me a browser, xterm, and quake and I can do 80-90% of all my computer needs.

  • ttmkfdir -o fonts.scale -p
    like the next post. It works better.
    you don't need to install any new rpms
    if you've installed RH6.0
  • The alternative to font anti-aliasing, in my experience, is a high resolution display, 1152x864 or higher. Font anti-aliasing was developed primarily for those people who don't bother to set their displays to anything other than 640x480. :)
  • >I for one am cheering for such a major upgrade.
    > With recent advances with Redhat 6.0,
    > GNOME 1.0, GTK+ 1.2, and many others (including
    > the anticipated release of Mozilla), Linux is
    > finally advancing far enough and fast enough to
    > be a serious contender in the desktop
    > market.

    a tip: from a recent convert to linux: make it suggest disk partitioning. the first time i installed it it was: "huh, partitioning? make it all one partition, of course."

    if the disk isn't partitioned, then *suggest* sensible alternatives. give advice. as a windows nerd of about 15 years, i found it a bit confusing.

    just my HKD 1.60
    dave
  • by Matts ( 1628 ) on Thursday June 03, 1999 @05:02AM (#1868978) Homepage
    And yet even with xfstt (or a patched XFree with ttf support) or xfsft (I've tried them all), font rendering at small point sizes is apalling. Absolutely terrible. No worse than PS font rendering at small sizes, but come on - Windows and MacOS have been doing good quality small point size rendering for years.

    I seriously hope this improves (it's all down to a good hinting engine). That and font smoothing would seriously improve my X experience.

    Matt.

    perl -e 'print scalar reverse q(\)-: ,hacker Perl another Just)'
  • wow.. really? Care to show me how to setup multi-headed Matrox Millenium G200 cards in linux?

    We tried getting them working at my job. Accellerated X said they could do it. Then they said they couldn't. (with those cards)

    Metro-X said they can do it.. then they said they couldn't. The only way to get multi-headed G200 support in Metro-X is by mixing a G200 with a Millenium or a Millenium II. You can't use 2 G200's.
  • Well, of course, Helvetica is the suckiest font that comes with X, IMO. Have you tried the 100dpi fonts yet?
  • Beeblebrox.

    Well, actually, multiple video cards and monitors. Don't know if it includes multiple keyboards and mice.

    Also don't know if anything other than a 1:1 card/monitor ratio is possible;ie, if any cards by themselves support multiple monitors.

    All of this implies independent monitors. I bet there are solutions already for monitors slaved to another monitor, but that's not multihead.

    --
  • XFree worked fine before (well, the font display sucked (yes, I know about standalone TT and PS fontservers)), but the design -- separate server binaries -- always bothered me as very inelegant. Finally, my aesthetic sensibilities will be satisfied, and we will all live happily ever after! (the fact that there will finally be native Voodoo3 support, won't hurt either).

    --

  • IMHO, other than hardware support, XFree is better. (And XFree has pretty good h/w support.)

    Reasons you might want a commercial X server:

    OpenGL with anything other than an NVidia or Matrox chipset. (And right now, 3D for the G200 is limited by incomplete specs.) So basically, h/w support...

    Multihead support. I'm not sure if XFree supports multihead yet. According to the announcement, 4.0 will.

    Umm... Other than multihead support and hardware support, I can't think of any other advantages... I think AccelX has been known to be faster for some cards, OTOH, I've heard many bad things about the quality of the server. (i.e. bugs) I'll take a performance hit for the stability of XFree.
  • It would take more than an addition to 4.0 to support that, it would take a pretty fundamental change in X-Windows itself.

    X fonts are returned to the server from the font server as monochrome pixmaps. Font servers expect to send that, x servers expect to get that, and the client programs all expect that. You'd need to extend the X protocol to support grayscale pixmaps for the fonts, recode the font server to be able to send them, and the clients to be able to understand them.

    IMHO, I don't see it happening. Some client programs where it would be useful could be extended to do it the way Gimp does it (request the font at a size a few multiples bigger than needed, and resize it down on the fly), but it would be application specific, and I can't see very many applications needing it.

    I thought for a while about poking around in Mozilla and trying to add the ability to do that in there, but since I decided to do it, I haven't managed to get a single copy of Mozilla to run on either of my decent Linux development systems...

    I think Mozilla is a place it'd be nice to see that support.
  • Well I for one would love to see support for antialiasing. I run 1280x1024 on a 17" monitor and notice a huge difference between text and graphics with and without antialiasing. I don't understand what the big resistance against it is all about. Just imagine if Linus, Raster or Miguel listened to the nay-sayers. Personally I think it would be a great addition to the capabilities of XFree86, and I'm surprised there's been no attempt to impliment it.

    Will the modular approach of 4.0 make such implimentation possible?

    Paul.
  • I would wager that xf86config can do it correctly
    as well.

    The riva 128 ZX is not really all that much closer to the tnt, certainly not in terms of 3d.
  • All things GNU are GPL

    I guess I should forgive you for not having a clue, since the link I posted doesn't work today. Try this link [gnu.org] instead. In short, not all of the GNU project is written by GNU, some of it is just included because it is good and it is free. And not all the stuff they collected from other places is GPL, though it is all free.

  • Though decent outline support would improve things no end -- I thought about implementing the double-sized fonts thingy in my X server, but gave up after taking a look around the innards of X (and yes, the most of the font system would need a rewrite, but it NEEDS one)

    Also, if you want a 14-point font at 75 dpi, then you DONT request the font at 28pt -- you request a 14pt font at 150 dpi (to be pedantic, but this show the problem about the font doubling -- you have to worry about the screen resolution, the font resolution etc. -- in short, and I think you'll agree, X can't do type to save its life).

  • If you are using Netscape to compare small fonts sizes, then you may be misled by the poor font quality on some web pages.

    It seems that Unix Netscape has a very old and well known bug that causes it to only use 12point fonts for scalable fonts. What this does is make very small fonts nasty and almost unreadable. You can only make a font so small before it is unreadable, no matter the system.

    What I do is when I start Netscape, immediately go to the preferences/fonts and type '16' in the font size box (this is assuming you are using a TTF font like Ariel, and have the 'scalable' box checked on.

    This makes things MUCH nicer and makes those small fonts behave.

    jf
  • A lot of good reasons and partition schemes mentioned above, but nobody seems to have mentioned a separate partition for swap.

    Sure, ideally you have enough RAM that you never swap. But if you are swapping, performance will be a heck of a lot better if the swap space has its own partition rather than messing with the filesystem. Some commercial databases like to have their own partition for data too, for similar reasons.

    On a mostly single-user machine (ie my personal workstation) I'm not much worried about overflowing /var or /tmp, so I generally have one partition for the relatively invariant stuff (the OS and apps, ie /var /tmp and /usr all on the same fs as /) and one for data, user files, etc. (Plus a swap). And then another partition for each OS if I want multi-boot.

  • Use the TrueType font patches, and install some TrueType fonts. it doesn't do anti-aliasing, but the scalability works well, try it.
  • Agreed, I have tried high resolutions (1600x1200), other fonts (URW) 75dpi, 100dpi. X fonts simply are ugly. Plus tampering too much with fonts makes some X apps (and their widgets) look wrong as they were not prepared for that.

    Some sort of intelligent font smoothing are really needed, anything else (as mentioned above) just doesn't quite cut it. Windows look crisp and clean with respect to fonts, in comparison.
  • > Not everyone needs, or even wants, anti-aliased text. For example, at very small point size, anti-aliasing tends to make things worse, not better.

    My car stalls when I start it in fourth gear. That's why I have three gears under that.

    You don't antialias small fonts. Windows doesn't antialias fonts under 8pt (you can make it, but it is awful). And has been said over and over and over and over and over again, it will require a new API.

    Or we could all use Berlin and get rid of the X monstrosity completely. Someone has to start.
  • # Font anti-aliasing was developed primarily for
    # those people who don't bother to set their
    # displays to anything other than 640x480.

    What about laptop users?!? 1152x864 is really not an option on most laptops.
  • by dwmw2 ( 82 )
    Will this new release support Xvideo?

    Yes. It supports the v4l devices and also I believe the Permedia inputs.

  • My Dell Inspiron 3000 runs Debian and X just fine
    at 800x600 at 24bpp. It's quite useable.
  • you can do whatever you want with it, as long as you don't remove the copyright notices

    Or the usage permissions

    This includes relicensing it under the GPL

    No, you can't really relicense it unless you are the copyright holder. If you make a new derived work that includes the old code from X plus some new code that is under the GPL. The result would only be distributable under the GPL. There's a clause in the X license:

    Except as contained in this notice, the name of the X Consortium shall not be used in advertising or otherwise to promote the sale, use or other dealings in this Software without prior written authorization from the X Consortium.
    That appears to conflict with the GPL's "no additional restrictions" clause, but apparently it doesn't, since the clause has no effect (even without the clause you would not be entitled to use the the name of the X Consortium). Somwhat unclear, that part, and many people prefer not to mix X-license and GPL code (it's also rather impolite towards the person who wrote the X-licensed code).

    At least the X license doesn't include the obnoxious BSD licensing clause [gnu.org] which is a pain in the backside whether or not you want to mix it with the GPL.

  • xfstt, the X11 fonts server for TrueType support uses the freetype [freetype.org] truetype rendering engine, which supports hinting.

    Look at the website [freetype.org] for more information.

    A part from the 'features' list on the web site:

    "A full-featured and efficient TrueType bytecode interpreter. The engine is able to produce excellent output at small point sizes. This component has been extremely difficult to get right, due to the ambiguous and misleadings TrueType specifications. However, we now match Windows and Mac qualities."

  • Windows was able to include it without any rewrite to existing apps. Are we actually admitting that Windows can do something that UN*X can't? And as for the need - try going to 90% of webpages that have hard-coded font sizes in them. The suckers get scaled down so small they look like little bit blocks. With anti-aliasing, at least I could make a guess at what it was supposed to say. And yes, I run at 1600x1200. It doesn't make much difference when people force their pages (or apps) to 6 point fonts.
  • They set up an 3D card so that you can map an X server output to the surface of a 3D object. You could have a cube with a program running on each face, or a wall with a program or programs mapped onto it. They're essentially moving the output of the server into the texture memory of the card. That should be feasible with the TNT cards, which can pack in a LOT of memory.

    Somewhat more interesting would be programs that could actually take advantage of the 3D environment. A conventional Xterm doesn't map well into that paradigm, I'm afraid. That would be an interesting field to research.

    With the recent advances in head mounted displays, you could probably also make your environment immersive. That would be cool.
  • -- Of course not. It's part of "the GNU system."

    That's funny, all the source states that it's
    still the property of the X consortium.
  • Posted by Scott Francis[Mechaman]:

    It's _stable_ but full of annoying bugs and lack of features. The MIT-SHM stuff was broken on my i740, leading to annoyings bits in Blender and a nightmare for painting things in the GIMP. And for some reason AX4.2 doesn't support DGA nor the ability to drop resolution below 640x480, which also didn't appeal to me. The i740 XFree module works almost as fast--the only difference I've seen is that it takes it about 5 seconds to display the root window on startup.
  • Where can I find a list of those 4 steps?
  • OK, time for me to come out of the closet-- I answer other's questions, but I don't know all.

    What's wrong with the one big partition approach? What's a better approach? What's most important to put on separate partitions?

    Generally, I use one big partition, or use the smaller drive for / and the bigger drive for /usr. Heck-- for fun, I might install devfs, and put / and /mnt on an initrd.

  • From my understanding, anti-aliasing would mean a substantial rewrite of many parts of the code, and would not be worthwhile. Plus, with a decent resolution, it really dosen't matter. There are even some people who claim that doing it makes it looks worse, not better.

    I am using xfstt as a truetype font server. both it and xfsft have worked well for me. And really, I don't see the need for anti-aliasing....it looks fine to me. i am happy that they are including native true-type support though.

    Can't wait till it's released. Thanks to everyone involved in this project :^)

  • FWIW, XFree has proven to be better than Sun's X-server under Solaris 7 x86 on our Ati RAGE IIC cards. Using the Sun version, it was sloooow and tended to leave pixels on the screen when scrolling text (the leftmost two pixels of 'm' characters would stay, for example). Switching to XFree has been relatively painless, except that CDE doesn't pick up the fonts for some locales (eg, en_GB.ISO8859-15). Performance has been rock solid.

    The only missing feature has been Display postscript; I didn't see it mentioned in the list of features, but I believe that work is being carried out on this.
    --

  • >Netscape tries to render fonts at a 12 point font base. This doesn't allow enough pixels to render the fonts correctly.

    Sorry. This should say "This doesn't allow enough pixels to render TINY fonts correctly"

    jf

  • We had a discussion about this very topic on a newsgroup a while back. My suggestion was to add an X extension that would basically say something like "Please antialias all of the fonts that I draw to this window" or "Please antialias all rendering with this particular font". At which point XDrawText etc. would have the same syntax and components, but the output would be antialiased, as requested.
  • I didn't see any mention of support for the Rage 128 chipset for 3.3.4 or 4.0 ... is there something I missed?
  • Glad to know I'm not the only one who remembers that horrible little line...
  • Jamie,

    XFree86 has been able to support multi-plane overlays for a while. When I was working on the Matrox Millennium driver, starting around August 96, the support got added eventually by one of the other dudes, um I can't remember if is was one of the other Andrews or Radek. It was easily two years ago that support was added - I haven't worked on the driver for a year now.

    There's just no easy way to make a different plane depth in X as it was shipped, and only some chipsets (eg, Matrox and a very _few_ others) can support it. I certainly never tested it, although I knew it was there.

    Andrew

  • The name Free86 was probably used because it sounds like 386.

    It's pretty irrelevant now since it runs on more than the x86 architecture.

    How about "FreedX" ?

    Just my stupid opinion.


    This sort of thing has cropped up before. And it has always been due to human error.

  • by Anonymous Coward
    Please let's make one thing clear:

    Since I read all the stuff about Linux in this comment: Xfree86 never was a "Linux project".
    (GNOME, GTK and Mozilla aren't either.)

    Yes, I know nobody said this but some readers here may get the assumption from reading comments like that.
  • The lack of True Type font support is the major thing that prevents me for totally migrating to Linux, Mozilla under Linux with it's ugly font really sucks. Yes I know, there is xftstt but I didn't manage to install it.

    With True Type font support Xfree86 will be a real contender in the Desktop realm
  • There is a set of nice type 1 fonts that you can use with XFree and ghostscript that suck much less.

    I think the name of the package was URW-fonts or something. It was available from www.gimp.org a while back.

    Using that, the standard X fonts look a lot cleaner.
  • It would be possible to get the X server to antialias text that is drawn with XDrawText(), but this would probably break some other applications (eg drawing the text again in the background colour to erase it may leave artifacts if antialiasing is used).

    That would definitely break things, and probably more than you think. There's no way to do that compatibly. To do this, you'd have to have a different API for drawing anti-aliasted text.

    Which isn't to say that adding that API would be a bad idea. I think it could be done in a way that was fairly convenient to use, and would degrade well on servers that didn't support scaled fonts.


  • The MacOS (or possibly even the hardware) handles the color bit change transparently - I don't think the application is even aware, because the system handles the dithering. Anyway, I used to have a IIfx with a 21" 4-bit grayscale and a 24-bit color monitor and everything worked (except a couple of games) with no slowdown.
    --
  • This is what we need to bring Linux to those that aren't even bright enough to use Winbloze.

    I personally like the idea since it's intuitive and makes use of that 3d card (since Im not much of a gamer). This will eventually come into play anyways (since they're already working on 3d displays and input devices) so why dont we get a jump on this area of computing?

    So why not guys? The first step would be a 3d version of XFree which maps 2d to 3d and a routine that maps the 2d mouse onto the top window and it's respective coordinate. This doesn't seem to far fetched since we've been doing it in so many games.
  • Um, no.
    Mulithead does not necessarily mean mutiple monitors driven by one card. In this case, Matrox cards will look for anothe video card at startup, and if it finds one it "turns off" its own "plain jane" VGA adresses. The system doesn't see these, and assumes its not a video card. After booting, if the X server supports it (MetroX does, AccelX does if you pay extra), it starts up another display attached to the system, and allows for you two have one keyboard/mouse and two monitors. If your mouse is on the left monitor, and you move it all the way to the right, it "jumps" to the right monitor and the focus goes accordingly. Each screen has it's own $DISPLAY variable, ":0.0" and :0.1", respectively.

    Until RedHat stopped including MetroX in the boxed set (at 5.1) I ran a dual monitor system with generic Trident card and a Millenium I. It had a dip switch on it to trun off the VGA adresses, but the Mill II's and up autodetect for another VGA card. If you have 2 or more Mill II/200/400's then the first one detected is the VGA console, and the others are blank until X starts.

    I was going to break down an get MetroX (and another 20" monitor, I already have the video cards), but if XFree 4.0 betas next month, I'll just wait till then. JOY!!!
  • That is because the X developers do not want to break old programs.

    Consider the case where you draw a string to a window. Then you want to erase the text without altering parts of the window outside of the text, so you draw the text again, but this time using the background colour.

    With X as it is at the moment, this works fine. After adding anti aliasing, you may get some artifacts after erasing the text.

    There were probably similar problems with windows using antialiased text.
  • try going to 90% of webpages that have hard-coded font sizes in them.
    That's why some of us access the web only through a clever filtering proxy [perl.com], which corrects all these design mistakes. :-)

    --tom

  • One of the main design goals of 4.0 is to provide a standard driver format so that vendors can write their own driver files, possibly in binary-only, which would be a drawback.

    ATI would have to write their own, or release specs so someone else could.

    jf
  • Which is the point -- it must be a option to be with, and an option to be without -- given the
    former, that latter is trivial, the X peoples'
    problem is the other way around.
  • Well, it was...

    Check out fsn on SGI's website. (Under "Serious Fun" then "Freeware", then "Software Development", I think)

  • Imagine a desktop with depth to push all those overlapping windows (er... wormholes?) away. Hang icons on walls. Turn your view for more space, and navigating on the fly :P

    It isn't hard if you try.

    X3DM: Instead of just catching up on the latest drag'n'drop features, offering the option of fly-or-fall.

    Seriously, I'm happy & productive enough with XFree+[a functional wm] as it is, but it's veery nice to read of inspirated development on areas such as multihead support - the one and only thing I miss in Macs. Inspirated, I consider, due to more than one line mentioning clarified design. Good [XFree developers' favorite item here]!
  • ehn...that hasn't always worked for me...I'm running 1600x1200, and Helvetica Bold looks kinda icky, though everything else is good...
  • by Anonymous Coward
    Is there a replacement for XF86Setup in the works, or are they looking for someone to work on this?

    I don't like Red Hat's text-based XConfigurator, and Sax looks nice, but it hangs my system. The current version of XF86Setup works the best, but it's pretty outdated.

  • The reason the small fonts do not look as good under X windows is because it doesn't anti alias them.

    It would be possible to get the X server to antialias text that is drawn with XDrawText(), but this would probably break some other applications (eg drawing the text again in the background colour to erase it may leave artifacts if antialiasing is used).

    By using the freetype library directly, or rendering to a pixmap then reducing the size of the pixmap, it is possible to do antialiased text, but it is difficult.

    The other option is to get a higher resolution monitor and run X in 100dpi mode :)
  • 'Since I read all the stuff about Linux in this comment: Xfree86 never was a "Linux project".'

    Of course not. It's part of "the GNU system."

  • by Anonymous Coward
    All this talk of how antialiasing can't be supported in X11. Too bad none of you have actually looked at the source or you would know it is in the XAA extension. Less talk and more coding would do you all good.
  • Similarly, XFree68 is that for M680x0-based machines (from the '68' in '680x0' [duh] and a word play on 'XFree86'). I suppose XFree21 would be for Alphas (kinda sounds like Century21, eh?).

    It would be absurd to call it by different names just because of the varying platforms, so why not give it a year number? XFree2000 anyone? (/me runs for cover).
  • Sure they *can* write binary only drivers, but I think the trend toward releasing open source drivers will continue. And when we have the choice, we (the Linux community) will simply go for the open source drivers. There's no reason to tolerate binary drivers given all the problems they present.

    I think hardware vendors will, in time, become cluefull in this area.
  • Note that the last line should read:

    "as well as 3Dsupport Precision Insight's DRI stuff (which looked really excellent at LinuxExpo), Mesa, AND SGI's GLX."

    Subtle but important for all of those who want/need to run GLX apps.


  • They may not have changed the name to just XFree yet, but they're set up xfree.org as an alias, so they're ready to do so if they so choose.
  • As well as having Multihead support, it also has the Xinerama extension. This allows you to join two screens to form a single display (a 2560x1024 display sounds nice :)
  • I hadn't heard of Xinerama before. I found a brief description about it on x.org [x.org]. It sounds fairly limited:
    15. Xinerama

    The Xinerama extension provides a way for a multi-headed system to function as one large screen. Windows can span multiple screens and can move from one screen to another.

    Currently, the Xinerama Extension works in a homogeneous graphics environment. A graphics environment is considered homogeneous if, for example, all of the graphics cards have 8 planes with 6 visuals. Mixing a 24-plane graphics card with a 8-plane card creates a heterogeneous environment.

    Unlike other multiple screen implementations, Xinerama provides a solution at the device-independent level. The advantage of this approach is that it reduces the amount of work involved in supporting and maintaining the extension. The number of graphics devices on the market continues to grow; embedding the extension functionality into the device dependent code for each device would be a maintenance nightmare. Since the Xinerama implementation does not require any low-level graphics modifications, existing device-dependent code does not have to be recompiled. In the loadable server world, the Xinerama Extension will work with existing device-dependent shared libraries.

    The Xinerama extension is not a standard. It is neither an X Consortium standard nor an X Project Team specification.

    I remember, years ago, being blown away when someone showed me a Macintosh with multiple monitors, one of which was a low-resolution 1-bit screen, and the other of which was a giant color screen. They dragged a window so that half was on one screen and half was on the other -- and both sides of the window displayed properly.

    I doubt X will ever be able to do that. I think it would require major protocol and API changes.

  • Yeah, I think geeks the world over cringed at that line, but it was only added for the movie.

    The book handles it much better - in the movie they added the interface and made the girl the unix-savvy one, for audience purposes.


    --
    make clean; make love --without-war
  • It's nice to see XFree86 moving along. It shows the strength of the open source model--just because the copyright holder has virtually abandoned the software, development continues.

    True, The Open Group recently announced some political reorganization of X, but as far as I'm aware, they haven't hired any engineers to work on it. (The old X project team left last summer.)


    Disclosure: I used to work for The Open Group, until last summer when they shut down operations in Cambridge, MA.
  • Uhm...X didn't have a standard UI until gnome? What kinda crack are you smoking, and where can i get some? First off gnome is not the standard X UI. There isn't one, and I don't think there ever will be one. Second, KDE predates GNOME by over 2 years. CDE predates KDE. And as for granting gtk the credit for giving X a ui, KDE runs on qt, CDE is motif, both of which aren't gtk and are in no way related. Also, XFree isn't following up on the example set by redhat and gnome. XFree has been around longer than either of them, and XFree is working on a stable 4.0 release, unlike the less than stable gnome 1.0 release. While redhat and gnome do deserve credit, it's no where near what you are giving them.
  • I've got a Hyperdata notebook here with an 8 meg ATI Rage 2 Pro 3D (those last four adjectives may be in the wrong order) but I can't get X to come up properly. The display is a nice 14.1" XGA and the computer itself is a P2-300.

    I've got the same card in my computer at home and it works fine. On the laptop, I can't seem to find the right refresh rates.

    My old laptop (Compaq Armada 1500DMT) worked great in X (800x600 mind you). I'd love to get this sucker working, if anyone could help. The laptop howtos don't cut it here. X starts up, it's just the refresh rates aren't right.
  • Regardless of how difficult it is to implement or how "bloated" it makes a design, the fact remains that anti-aliasing is necessary to produce a pleasant display. Ditto the rendering of fonts at small point sizes.

    If you assume that what is necessary is inevitable, a fair assumption, then you can provide useful feedback. So:

    What's the cleanest way to implement anti-aliased font rendering? Alpha bitmaps? Which side of the client-server connection should it reside on? Can this be done without breaking any clients?

  • FWIW, you don't need _outdated_ Matrox cards, all Matrox cards (except the Mystique) support multihead.
  • I personally see nothing much wrong with implementing anti-aliased fonts, not implementing them as an extension (the creeping featurism can easily be solved by removing most of the legacy X font support and placing it into a demand loaded module). The rewriting of applications talked about is less than people say, since modifying GTK and Qt alone would do most of the work for most applications (though Motif maybe a lost cause (in this respect -- it's a lost cause in general anyhow)). That said, does anyone have any pointers to discussions detailing how exactly AA would break applications? Also, for use on laptops, adding a 19" 1280x1024 monitor isn't really an option -- and being able to use both AA and sub-pixel AA would be useful -- the ability to have AA should be considered far more important than actually having AA (think maximal generality) such that it could be added and removed transparently to the applications using the display. Given that UN*X is known for its tendency to have its tools used in ways not originally designed for them (e.g. using a tar-tar pipe to recursively copy a directory and preserve the file dates), ruling a feature out becuase "I can just get a higher res display" seems to be very short sighted.
  • >640x480 is just a PITA with X, for reasons other >than fonts. True -- X has a reputation for inflexibility. (Why does it ALWAYS have to work at pixel level?)
  • Actually, you don't really want AA in this case, you only want AA to make 10+ pt fonts look better. With your microfonted web page, instead of looking like a bunch of dots, it would look like a smear. Quite an improvement I'm sure.
  • It has nothing to do with antialiasing. I'm using Windows, I have to support it in a technical support environment, and I can tell you that I hate antialiasing. It just makes everything soft, blurry and kills all the contrast. Windows' rendering of TT fonts is, for the mostpart pixel-perfect. There must be a trick to get this stuff right in the rendering engine or something. It is as though someone manually mapped each pixel in the character cells. Right down to little dips in the Roman letter 'c' to 45 degree angles in the letters 'k' and 'x'. I think there is more intellgence in the engine than an automatically scaled and generated character can provide.

    Eventually, Xwindow will have to provide similar clear, sharp, efficient fonts. I find X difficult to use simply because either the fonts are too big and surrounded by copious whitespace, or the fonts are so crushed and small that the letter 'e' looks like a reversed apostrophe or a smudged 'c', and the letters k, d or p look like a chunk of burnt meat on a stick.

    I really would like Xwindow to be more pleasant. but antialiasing is a cheap hack. Given the choice, I'll take meat on a stick over fuzzy meat on a stick any day.

    It could be sooo much better... and I know there must be a way to fix it, but it shouldn't ship this ugly and unusable.

    Some day I might be able to do better and contribute. Alas, this is off topic... Just ignore me.

  • do this as root:

    get the fonts from wherever. /mnt/c/windows/fonts/ in my case...

    cp *.ttf /usr/local/share/fonts/ttf/
    (or wherever - make sure to create the directory, if it doesn't exist)

    cd /usr/local/share/fonts/ttf/
    (go to the directory)

    ttmkfdir -o fonts.dir
    (make the font index file)

    chkfontpath --add /usr/local/share/fonts/ttf
    (add the directory to the font path)

    /etc/rc.d/init.d/xfs restart
    (restart the font server)

    There. All done. Use chkfontpath --list to see your current fontpath, chkfontpath --remove to remove entries. Try chkfontpath --help for a summary

  • Xfree86 3.9 when I was working on it (from Aug 96 - May 98) used a great deal less memory than XFree86 3.3.x. That's because it only loads the chipset, color and depth support you're currently using. I don't know about memory leaks, but if you're really worried about it, why not join the effort and help out?

  • Matrox has a few cards that do multiple monitors on one card. There are 2-monitor and 4-monitor G100 cards, and the G400 will have really nice built-in dual-head support.
  • Will this new release support Xvideo? I remember Metrolink supports this in their Xservers, and the current release of XFree doesn't. Perhaps they call it by another name...
  • there is a product like this for win95/98 at www.3dtop.com. I havn't played with it in depth since It dosn't render full screen so can't really take advantave of 3dfx cards unless i used winglide and iv'e been too lazy to get it working. It dosn't look all that impressive yet but with various plugins you can load 3ds models as icons etc etc.
  • Most things that makes Linux good are not really
    Linux projects...

    And that's the way it should be...

    Xfree86 definitely is a big part if the system
    that most people refer to when talking about Linux...

    /Daniel
  • Precision Insight and SuSE are building in the direct rendering stuff, which should bring OpenGL up to Windows level and make it easier to support 3D in general
  • Make anti-aliasing an option that has to be explictly turned on for given window. Then old apps will continue to function, and new apps that need it (web browsers, word processers, other 'pretty text' programs) can turn it on. And new apps that don't need it can keep the speed of not antialiasing.

    Personally, I've never had a problem with X's text anyway, so I could care less. I don't think much, and never have, of true type fonts. Adobe Type 1 fonts work fine for me. But I think the above would work for those who need/want antialiased fonts.

  • I assume Xinerama lets you drag windows across multihead boundaries. (I also assume that without Xinerama you have to start things on different heads DISPLAY :0 or :1)

    Should Xinerama work with two different display resolutions? I eventually want to run 1600x1200 on my big head and xga on my little head/s?

    Where can I get more info on Xinerama?

    Ed
  • I'm glad that by the end of the year we'll have true multihead support, without buying MetroX and two outdated Matrox cards. Well, I'm assuming XFree will be able to do multiheaded with any of its regularly supported cards.

    -NG


    +--
    Given infinite time, 100 monkeys could type out the complete works of Shakespeare.
  • I disagree that anti-alasing is needed. I have netscape using tt-fonts and it looks absolutely beautiful. Well smaller fonts look like ass but, eh, it happens.
  • How does XFree86 compare with other commercial X-servers? I am thinking about features, not hardware-support.

    // Simon
  • I guess you haven't heard of the TrueType patches for XFree86 - I used them, all you have to do is follow the instructions to patch it in, rebuild, and get a utility to build a fonts.scale file for TypeType fonts (there's such a utility listed on freshmeat). Mind you, you'll need a few hours (took 3-4 hours on my P100), but it builds TrueType support (via the FreeType library) into the X server and the standard xfs fontserver (and mind you, a fontserver is actually frequently a better way to go - it can keep the X server from blocking while fonts are being rendered - hopefully XFree86 4 will use threading in the X server itself, and make this less of an issue).
  • I'm hoping that they include support LCDs attached to the ATI Rage Pro LT (the current server doesn't). Then I could finally use my Inspiron 7000 with the swank 15" screen. There's a bunch of kludges to get it working now, but they're broken under the most recent BIOS. And I need the most recent BIOS for good WinNT support and to test Win2K to see how much it sucks.

    Waaah.
  • Unless I am mistaken (could be) the 86 in XF86 is from the x86 (ia32 intel) architechture. Now that Xfree 86 supports many architechtures isn't it about time to dropthe 86 from the name?
    "There is no spoon" - Neo, The Matrix
    "SPOOOOOOOOON!" - The Tick, The Tick
  • by timur ( 2029 ) on Thursday June 03, 1999 @06:41AM (#1869081)
    What makes TT fonts look good at small point sizes is "hinting", not anti-aliasing. AA makes medium to larger fonts look better by blurring the edges so that they appear smooth. Hinting is a set of instructions inside the font that tell the font rendering engine what modifications to make to the font at small point sizes. Microsoft's TT website has quite a bit of info on it - see http://www.microsoft.com/ty pography/hinting/hinting.htm [microsoft.com].

    Given the same TT font on both X and Windows, if X shows the small points worse than Windows does, then my guess would be that the hinting support in X is either missing, broken, or just not good enough.

    --
    Timur Tabi
    Remove "nospam_" from email address

  • It would take more than an addition to 4.0 to support that, it would take a pretty fundamental change in X-Windows itself.

    I don't think so.

    X fonts are returned to the server from the font server as monochrome pixmaps. Font servers expect to send that, x servers expect to get that, and the client programs all expect that. You'd need to extend the X protocol to support grayscale pixmaps for the fonts, recode the font server to be able to send them, and the clients to be able to understand them.

    You could do all of this on the client side:

    • You want to draw 14-point anti-aliased text.
    • You request a 28-point version of your font.
    • You render your text to a scratch bitmap.
    • You dither it down to half that resolution.
    • You display that pixmap.

    Now obviously that's pretty inefficient, and this could be done much faster on the server side, if the server had support for it. But the basic mechanism would be the same, it's just that instead of using a scratch pixmap, the server would do the blending directly onto the target drawable.

    So it's easy to imagine a system where the ``draw an antialiased string'' function would do this negotiation behind the scenes: if the server supported the right extension, it could let the server do it, otherwise, it could do it the hard way. (It could even load the double-sized font for you, by looking at the font you've passed in, and keeping a cache of double-sized versions.)

  • The way I remember after reading about this, adding anti-aliasing would break compatibility with current X programmes, so that the new feature would have to be coded in as an 'extension' of sorts, and that would require that current X clients be rewritten to use that extension. This would either increase the creeping featurism and bloatiness that is currently present in X. Besides, as you say, it's not really needed.
  • I don't see that the X consortium are going to add anythin in the near future -- if X is going to have any chance (though I hope that an alternative takes over personally) is for XFree and the commercial vendors to get together and do it themselves

  • On deja news (which should nearly always be your first stop for answers to linux questions that aren't in any of the pieces of standard documentation to your knowledge).

    Here's a relevant post:

    1. install the package freetype-1.2-6.i386.rpm
    2. then cd to where your TrueTypes fonts are and /usr/sbin/ttmkfdir -o fonts.scale -p
    3. '/usr/sbin/chkfontpath --add [dir]', where [dir] is the name of the dir with the true type fonts
    4. /etc/rc.d/init.d/xfs restart (don't xfs stop and xfs start, it messes up running X sessions awful) and you're ready to go

    (based on a deja news post from Alexandre Blanchette

  • TrueType fonts are great, but anti-aliased fonts are more important (IMHO). Does anybody know if this is part of 4.0?

  • Netscape (4.06 and 4.51, anyway) does save font sizes, just like it does all the other preferences. Maybe you've got a permissions problem?
  • As far as I'm concerned, this is the best news in their announcement:

    • 8-bit PseudoColor overlays when running in a TrueColor mode (on selected hardware).

    This is something SGI has done forever, and it's just incredibly convenient. It's so much nicer to be able to have the default visual be an 8-bit colormapped visual, but have it be possible for specific applications to use 24-bit color as needed. Most applications don't need TrueColor, so all that memory is just wasted on them. And there are things you can do in PseudoColor that are just impossible to do efficiently in TrueColor.

    It also makes debugging X programs much easier, because you can test whether your application works in both PseudoColor and TrueColor modes without having to start a second X server to do it.

    I wish they wouldn't call this ``overlays'', though. Overlays are something else entirely (that's the term for visuals that have transparency built in at the hardware level. That kind of thing is supported on X servers from all the major players except XFree and Sun: SGI, HP, DEC and IBM.)

Those who can, do; those who can't, write. Those who can't write work for the Bell Labs Record.

Working...