[syslinux] Syslinux direction

Jared Rhine jared at wordzoo.com
Sat Feb 26 02:13:55 PST 2005


I just joined this list, though I'm a long-time syslinux user.  I read
the archives before joining and saw HPA's thread on "syslinux vs
grub".  As he solicited feedback, I'll give mine here.

I use both syslinux and grub, in somewhat different situations.  I
appreciate both syslinux's clarity and grub's features.

For machines booting off hard disk, I generally use grub, because when
problems arise, I want as many features as possible to assist in
recovery.  Syslinux doesn't, and shouldn't, compare to all that
command-line crap you can do in grub.  This is less and less important
these days though because for recovery, I now boot into a live cd, not
grub.

For network boots, I use pxelinux exclusively.

For CDs, I used to use syslinux always, but I've built more and more
CDs using GRUB.  The reason has been the menu system.  It's pleasant
to use the up and down arrows and to be able to edit a boot line.
It's pretty, mainly in the ease of setting colors, and the box around
the menu itself.  (Every GRUB-based CD I make though includes a copy
of memdisk, though :)

With the inclusion of simple menuing in syslinux 3, I've got to
revisit the situation and see how that stacks up; I frankly sort of
missed that development.  If it gives me the basics done well, I'll
probably stop using grub on CDs.  I'm very unlikely to ever use the
advanced menuing.

I believe there's a clear market for syslinux.  The market for boot
loaders is very small, and the grub project looks and works just like
a standard GNU-sponsored project, and that's not a real complement.
(Cranky developers, the "GRUB Legacy" thing, etc)

Syslinux has a clear lead, and can continue to dominate the CD,
networking booting, and other "custom distribution" areas.  I think
GRUB is probably going to stay competitive in the standard "boot from
a hard disk" which makes up the majority of actual boots out there.

In general, for my purposes, the features that keep grub in my toolbox
are:

  - Menuing (noting above that I've got to learn syslinux menuing)

  - Multiboot support (Required for Xen boots; you asked earlier what
    technically is going on in Xen and why syslinux doesn't work; the
    answer is that it uses the multiboot spec to have a place for both
    the ELF Xen hypervisor, and a module for the Linux kernel itself.
    Someone said they didn't think any released live CD used GRUB,
    well, my XenCD does for this very reason.)

  - Ease of configuration (single file, not a separate file for
    colors, for instance.  Ease of configuration matters for market
    penetration purposes, including menuing.  If more of advanced
    menuing was "built-in", it'd be a clear advantage; I recognize the
    annoying complexity and bloat that would go along with that,
    though.)

If I had to ask for a direction, I'd ask for 1) preserve the progress
of the project, don't go into maintenance mode, 2) multiboot support,
solely for XenCD, 3) incremental improvements to the configuration
file, menuing, keyboard navigation, etc, 4) a focus on "non-hard-disk"
use cases if effort needed to be budgeted.

If there's confusion about pros and cons, position within the market,
etc, it might be helpful to prepare a comparison matrix, even post it
on the syslinux site for the benefit of the community.  Lay out
exactly where grub is better, further dividing it into which items you
care about and which you don't.

Hope this was helpful even though the original thread is cold.  Thanks
for all the work!  Within syslinux, the whole live cd movement (and
it's a big movement) would be less than it is today, since so few
people understand how to do booting right!

-- jared at wordzoo.com

"Truth is a great flirt." -- Franz Liszt




More information about the Syslinux mailing list