Wednesday, July 26, 2017

TenFourFox FPR2b1 available (plus: come to VCF!)

TenFourFox Feature Parity Release 2 beta 1 is now available (downloads, hashes, release notes). Besides various and sundry additional optimizations once again shamelessly backported from the Quantum project, this release includes the accelerated AltiVec string matching routine which I will be also hooking up to other parts of the browser in future releases, and reworks the font blacklist so that Apple sites work again (and more efficiently). Release is planned for August 8.

With this beta you will notice I've made a new build available - a true "Debug" build. Since currently we really only have comprehensive test coverage for JavaScript, this gives people a chance to help debug other browser issues without having to build the entire application themselves. I intend to only issue these for beta releases since the idea is to have as few changes as possible between the beta and the release version. Please take heed of this warning: the Debug version is not for mere mortals. Because it has a full symbol table, it could cause your system's crash dump facility to hang if it bombs and you weren't running it within the TenFourFox debugger (and if that happens, you'll have to kill the crash dump or it will peg your CPU; I actually hacked /usr/libexec/crashdump to not run if I've touched /tmp/no_crashdump as a flag). In addition, the debug build is much slower, runs a lot of sanity checking code and generates copious logging output. It is not optimized for any processor and has no AltiVec code, so it will run (badly) on any compatible Power Mac. Please refer to the developer instructions and understand what you're doing before you run it, and if you do run it, I strongly suggest creating a separate profile specifically for debugging if you intend to work with it often. The debugging versions will not be offered from the main TenFourFox page so clueless folks don't grab it by accident.

Next up, FPR3 will have some additional optimizations and performance improvements too, but will be focused instead more towards supporting new features again: in addition to some minor compatibility tweaks, I would like to get enough of our CSS grid support working to be credible and do further improvements to our JavaScript ES6 implementation. More on that a little later.

A schedule note: I'll be demonstrating the Apple Network Server 500 (the original floodgap.com) at this year's Vintage Computer Festival West August 5 and 6 at the Computer History Museum in Mountain View, CA (last year's exhibit was good fun). The ANS (codenamed "Shiner" after a local brand of beer in Texas, where it was developed) was arguably Apple's first true-UNIX server (A/UX notwithstanding), almost certainly the biggest computer they've ever made, and the last general purpose computer they built that wasn't a Mac. I ran mine almost non-stop from 1998 to 2012 as my main server (14 years), and it came back briefly in 2014 when the POWER6 currently running Floodgap blew its mainboard. I'll have it set up so you can play with it, plus an actual prototype Shiner for display and a couple (haven't decided?) PowerBook clients to demonstrate its unusual software powers. Be there or be less nerdy than I am.

Oh, and at last, Flash is dead (by 2020). But it was dead to us in Power Mac-land a long time ago. Finally the rest of the world catches up.

5 comments:

  1. Too Bad the HTML5/EME to play Netflix and other DRM for Hulu isn't open source or compiled for PowerPC. Wasn't V8 going to be compiled for Power8 for IBM? IDK if that would help.

    ReplyDelete
    Replies
    1. V8 was (not sure how current it is), but it won't really help here because we can't run the binary DRM blobs.

      Delete
  2. When I trigger the bug from issue 72 in the debug build I get myriads of this in the console:
    2017/07/27 22:08:40 [0x0-0x1e61e6].com.floodgap.tenfourfoxdebug trying to deliver this event too
    Nothing in the browser console. Anything else to look for?

    ReplyDelete
    Replies
    1. Damn it, I was afraid you would mention 72 isn't fixed. :P I couldn't repro it on my 10.5.8 DLSD, but again, I've had trouble reproducing it at all.

      However, that message is actually very helpful. It says that there is an orphan event with no target, similar to what happened with old issue 248 (though that could be triggered on 10.4 too after some work). With the console open, try clicking on the menu bar after the issue triggers. Do you still get the same message?

      Delete
    2. Actually, I get these messages in the console as soon as I start doing the STR. When I then click on the (non-functional) menu bar, I get another (smaller) load of these messages. See log at issue 72.

      Delete

Due to an increased frequency of spam, comments are now subject to moderation.