Hi all... I'm looking once again at the Syslinux roadmap. It seems to me that we're going to have a very hard time making Syslinux 4.00 be the full-featured everything-in-C release I had hoped for, or perhaps more specifically, that it is justifiable to sit on the current filesystems-in-C code waiting for the rest. That would probably mean productizing the transitional COM32R module format, unless a better solution for memory management can be found. Looking forward, I have experimented with using lwIP for PXELINUX. The good news is that the performance is *very* good -- I have reliably pulled about 250 Mbps over a local network, with peaks up to 600 Mbps. The downside is that it requires architectural changes which may make it *very* hard to support COM16 modules going forward. This will obviously break Gfxboot. As such, I'm looking at revising the roadmap, and I'm partly asking for input what it should look like. One possibility is: Syslinux 4 - Filesystems in C, COM32R, basic interfaces to support Gfxboot as COM16. Syslinux 5 - ELF linker, native TCP for PXELINUX, no COM16 support. Syslinux 6 (or 5.something) - EFI support. I'd like to get people's feel for the timeframes they care about these features for... -hpa
H. Peter Anvin a ?crit : [...]> As such, I'm looking at revising the roadmap, and I'm partly asking for > input what it should look like. One possibility is: > > Syslinux 4 - Filesystems in C, COM32R, basic interfaces to support > Gfxboot as COM16. > > Syslinux 5 - ELF linker, native TCP for PXELINUX, no COM16 support. > > Syslinux 6 (or 5.something) - EFI support. > > I'd like to get people's feel for the timeframes they care about these > features for... >Hey Peter, That roadmap is very interesting to read. I was just wondering if we could add LUA somewhere. LUA will allow interesting features as "hardware related booting". Maybe we can add it in Syslinux 5 not to load too much syslinux 4. I'm backing you in delaying the com16 deletion. That will give more time to adapt gfxboot and existing modules. I can't hide my happiness to see the Lwip incoming in Syslinux 5. HDT will have the benfits of it to export reports ;) About the timeframe, on my side, I don't really care about it. That's a free software, no one get paid for doing it. That said, having this kind of roadmap is nice as it give a realistic list of content to achieve before releasing. Cheers,
On Thu, 15 Oct 2009, H. Peter Anvin wrote:> As such, I'm looking at revising the roadmap, and I'm partly asking for > input what it should look like. One possibility is: > > Syslinux 4 - Filesystems in C, COM32R, basic interfaces to support > Gfxboot as COM16. > > Syslinux 5 - ELF linker, native TCP for PXELINUX, no COM16 support. > > Syslinux 6 (or 5.something) - EFI support. > > I'd like to get people's feel for the timeframes they care about these > features for...I've looked into it a bit and it looks like turning gfxboot into a com32 module (with the current api) is not a problem. It will need a big enough low mem bounce buffer, though. This should get us going until EFI (I hope to have the C-rewrite ready then). I'll get the com32 module a try and then report back. Steffen