Dirk Nehring dixit: >I follow Phil's opion: don't drop brcm-2.6
Purely technically (yes, nit-picking this time), we can't drop it because we don't support it now anyway. How about the brcm-2.6 discussion will be postponed until after 1.1 is released? There is much other stuff to do, can you spell nfotiser? >view, we are waiting since months for a working brcm-2.6 plattform: One which doesn't output anything on the console right now. >better wireless support (in theory, since 2.6.22 with OS-drivers), Drivers that can barely do anything (well, the BLOB can't do much more) for hardware that ought to be replaced by cards with other chipsets, true. >current kernel development with many needed features , bugs and instabilities. To cite from the diner at LT: The Linux kernel maintainers think it's the duty of glibc to provide a stable API/ABI to the user and they're thusly free to break things whenever they want. Also, they don't care about regressions at all. >gcc4 http://blog.fefe.de/?ts=b8af0295 I've now (for BSD) finally decided that we will keep gcc 3.4 exclusively in-tree as long as it's possible (for several years at least), and compile everything with it where it's still possible, and gcc4 is only provided as an optional alternative. Too much existing code breaks with it. Language lawyers, such as Andrew Pinski, hm let me phrase it like this: “ Those who give up reliability for a little benchmark-only performance deserve neither reliability nor real-world performance ” (free after Benjamin Franklin). I've been monitoring gcc progress for a while (since 2.7.2.3 I think); yesterday, I coerced the SUNpro C Compiler from SunStudio 12 (for Linux) to build both Linux and native MirBSD binaries _in the emulation_ (and found an optimiser bug). Too bad this one only supports sparc and i386. Of course, we should offer gcc4. Some software _has_ fixed all the am- biguities in their source code and really benefits from it. Some plat- forms do, too (most notably amd64 – I'm not really sure about sparc64 wirh _any_ version of gcc, Tonnerre Lombard knows the ugly details). But if we do, we should consider making it a per-package option instead of a global option. We already have the framework (honour $CC, $CFLAGS, etc.) – now you know why I did that. >Phil is right here. I would like to see even more support for other >hardware like the FRITZ!Box series (great VoIP plattform). I personally would like to see support for the Mips Malta board, even if it's into thousands of US$ price-wise. Why? Because that's the most complete MIPS platform qemu emulates. Why? Now go and read in what area I've committed yesterday. Read the do-build target. You'll understand. >Waldemar, I hopefully understand your point. Hm, _now_ I'm lost. Did I miss something? What did he want to say then? >* prebuild, complete images (well defined feature set needed by most of > the folks) No. We have the WIB. We can extend it with some kind of profiles. (Well, I think Dave can. I don't know Ruby, without or with rails.) >* a web interface (oh yes, everyone wants it, only developers hates it) It's in the makes. You should come on to IRC more often ☺ Ralph Passgang dixit: >Am Dienstag, 5. Juni 2007 02:55:00 schrieb Thorsten Glaser: >> Ralph Passgang dixit: >> >> Furthermore we have a nice OpenWrt release, which everyone can use, if >> >> somebody needs Broadcom Routers with 2.6! >> > >> >Waldemar, is that really you writing this? :) >> >> I think both sides just have drunk too much… > >I cannot talk for waldemar, but I haven't drunk a "bit" *fg* I meant him and OpenWrt. >the webgui Hmm. >the ifup-rework Okay, I missed that. Apologies. >routerboard-port That's been existing for a long time. >I think he wanted to start a discussion Oh no, let's just hack and improve it, 'kay? >I also think it's free enough already (and for the actual size of the >project), but I also wanted to show up some possible alternatives (but more >for the future in general). You can't compare a project with 10 developers >with another project with 100 or even 1000 developers. They have diffrent >needs in their "political" structure. I *so* second that. >> Elections? *laugh* > >it works at least for debian since over 10 years and over 1000 of developers. >It's not the badest of all possibilites. Sure, but we're not Debian. >But don't get me wrong. I don't want to change a thing in freewrt, I just >wanted to correct the "never change a running system" statement. If you It's okay. >really would believe in this statement, then you wouldn't tried stuff like >removing the +x bit for init.d scripts, would you ;-P Er… let's talk about something else. Markus Wigge dixit: >Bad idea I think. Lets focus on getting 2.6 running and drop 2.4. No, not for 1.1 – except rb-2.6, which seems to be the platform of choice on Routerboard for what I've heard. As far as I'm informed, Linux 2.4 requires much less resources too, so we shouldn't drop it as long as we support devices with either less than 8 MiB flash or less than 32 MiB RAM. (IMHO, YMMV) >This read like "We only will support Routerboard in the future" and that >is another bad idea. *agrees* >2.4 is dead for years in my opinion. Not now, but in a few years. How about we either don't discuss that right now, or something like, drop 2.4 support for the first major release after upstream drops development of it, unless some kind of MIPS community takes over maintainership, or if it's broken or the devices no longer available, or something like that happens. >Well then, let's first of all focus on the hardware available at the >market. At least those two Asus routers and the G3G Linksys. “La Fonera”. Be well, //mirabile -- I believe no one can invent an algorithm. One just happens to hit upon it when God enlightens him. Or only God invents algorithms, we merely copy them. If you don't believe in God, just consider God as Nature if you won't deny existence. -- Coywolf Qi Hunt _______________________________________________ freewrt-developers mailing list [email protected] https://www.freewrt.org/lists/listinfo/freewrt-developers
