So here is the idea I had.
Can we produce a "really light" version of VL that would be aimed @ even older boxes than STD is?
I support this idea. At the time VL 5.1 was being developed, I commented on the system requirements being slightly more than what was typically available in machines found by "dumpster diving". Even now, there are a lot of discarded Win98 or early XP machines which have 300-800MHz processors but only 32-96Mb RAM. SIMM memory modules of 32Mb can be had for next to nothing, but larger ones are rarely available used and are not particularly good values when purchased new. The fact that most motherboards only offered three memory slots means that populating a board with more than 96Mb of RAM exceeds the point of diminishing returns.
There IS a "market" for a GNU/Linux which targets resurrected machines and I think that Vector should offer a version which targets that market. The reason to do so is to encourage further development of Vector. Not only would you gain the ambassadorship of users receiving machines distributed from people like Lagnanon (who refurbishes older machines for the disadvantaged), but current VL users might be more likely to resurrect an old freebie than to risk their understandably treasured primary installation for purposes of experimentation.
The goal of attracting new
developers (not to mention new users) by offering a basically "no-cost" platform upon which to experiment should make MOE-lnx's idea well worth pursuing. I, myself, have had little interest in contributing to Vector development but would be very interested in dedicating some time towards the goal of producing a minimalist version.
Here is a a couple of things I have in mind
* Keep the ISO under 300mb (DSL is only 50mb)
* Ship JWM and FLux (configured like in 5.9std)
This ISO would only contain the most basic but useful applications
* One browser (make it dillo to keep it light)
* one good terminal app
* light Office apps (gnumeric, abiword (or even other lighter ones))
* No kernel src
I agree with you other than I think Dillo (as much as I like it) is a little too barebones and an additional option should be offered. Also, terminal apps typically consume only tens of kilobytes on a CD and little is gained by limiting available options (though if the default menu setup only presented one terminal, that would be fine).
Likewise, there are quite a few window managers available which require well under a megabyte of disk space and I would suggest that if a developer volunteers to package and maintain one of these, it should be a valid candidate for inclusion (for myself, I would offer to maintain a UWM package requiring 50Kb on the CD). Again, the default installation could focus on just one or two, but I see nothing wrong with the CD including a couple of megabytes of lightweight WMs.
One thing that should need to be addressed in creating a ultralight VL version is a refactoring of the Vector bulk tarballs. The current bulk tarball for core functionality (no X11) is 80Mb, which seems to be a somewhat bloated starting point for minimalist version.
To be honest, I wonder if it might not be prudent to reconsider the utility of the bulk tarball approach to installation. Back when I first used Vector (version 2.0), the space savings (on the CD image) by the bzipped bulk versus the original gzipped Slackware packages was significant -- with the advent of Vector packages being offered with LZMA compression, I fail to see a benefit to bulk tarball installation methodology. It would seem to me that a return to individual package installs, with the appropriate use of SW's tagfiles, would provide some welcome flexibility to installers (custom and repeatable installations, network installs, installs from non-standard devices).
Is there some benefit offered by bulk tarballs which I am missing?
