--- Begin Message ---Second, we may have to compile a different kernel for VL modem installs. Carl may have to drop back to the 2.4.18 kernel to get the Intel HaM driver to work. He also believes it is worth investigating the 2.6 kernel. The 2.6 kernel should run faster on a low resource box than 2.4.
- From: Richard Zimmerman <csxt@xxxxxxxxxxx>
- To: frgeek-michiana@xxxxxxxxxxxxx
- Date: Sat, 03 Apr 2004 00:46:44 -0500
Errr.... I downloaded and installed 2.6.3 on my smp machine... The initrd file was SIGNIFICANTLY LARGER in 2.6.3 verse 2.4.24! (read, it's huge) The chances also are the 2.6 will require more memory then these limited machines have @ 32mb. Also, how much benefit would a FGM box gain in having a 2.6 kernel? To keep the project in focus, our clients are not going to be power users like we are.
Third, Carl brought up devising our own distro. I almost choked on that one. If we use a 2.6 kernel, we are also looking at recompiling apps. At that point we might as well roll our own distro, which brings up long term support and maintenance issues. Carl favors Fedora as a base. I almost choked again.
This is not a good idea . We can not cover all our basis now without the added commitments or rolling our own Distro. Yes I played with rolling my own distro based on HR 7.3 and 1. It's *NOT* an easy task and 2. It gets cumbersum quickly. If we even think about it, I'd strongly recommend a lighter weight Debian Distro. At least then our clients can install their own software upgrades easily.
I also recall reading something today about FGP maybe working with someone to roll a distro. If we truly are working towards that, we might want to join forces to spread the workload.
Fourth, we may want to test purchase other inexpensive modems which are better supported for the Linux 2.4.20 kernel. Jay suggested purchasing used hardware modems (internal) for $15 at surplus. Unfortunately most of them are 28.8 kbps.
I agree with Jay on this one. Hardware based modems are the way to go if we can get them fast enough. If not, we need to keep working on the present Intel Ham Driver.
Goose
--- End Message ---
