Dominique Leuenberger a.k.a DimStar wrote:
Well, as you have such great scripts and the all-around testing experience:
Why don't you just package your script up, submit it to Factory and change the integration part of Yast / Bootloader config to use your scripts that does everything and more, just without error?
If I thought I could write something I would find acceptable for submission, I would, but my standards are such that that I rarely think my code would be worthwhile for others. The code I write that is reliable enough is usually too specific -- i.e. it's reliable because the specification for how it needs to perform is sufficiently narrow that it can do the task well. I'm very aware of the issues in supporting complex software. While software complexity rises geometrically, if one is lucky, testing doesn't shrink -- I say lucky, as testing needs to rise in proportion to complexity. But usually, the contrary happens, and major companies release product and call it Beta quality -- and that's become the standard.
THEN, and ONLY then, when NOBODY has any issues with your solution, can you make claims that your solution is so much superior to the existing perl-bootloader soltution.
I make no claim of superiority. The discussion was centered on the problem of having to call lilo as part of a kernel install. I and others pointed out that that by using a script (including post-install scripts in rpm), one can do just about anything one needs to support their installation. Someone posted their script that worked with grub for a 'hand install' (one could argue about that definition endlessly when it somes to the fact that it's all done in electrons). As for perl-bootloader vs. my kernel installer? well I make no claims of superiority, but on my system, mine works, and perl-bootloader doesn't. So, I don't really care if it is better -- it first has to work before better is even considered. When the suse-boot-installer USED to work -- it took ~3 minutes to run, most of which was spent building a RAM disk image that I don't boot from. vs. my script ... that installs kernel + modules in <1 minute. Used to be about 20-30 seconds, but kernel images have gotten a bit bigger and lilo makes a map of every sector of every kernel image. My script broke when lilo broke for a few months and suse no longer supported it... fortunately the lilo maintainer was already on it (kernel got too big) for certain sector maps... Other than that... it's worked fine for ~8+ years. But I make no claims about general usefulness, nor have I tried to make it general. It's not worth the effort. However, if even if it was slow -- the fact that it works make it better for my situation than something that doesn't. Nothing personal, but I prefer running to not? ;-) -l
Think about more complexity than what your system offers...
Point being: openSUSE is the distribution produced by the openSUSE community. SUSE happens to be an important 'player' in this community, but does not have the overall responsibility... when oh WHEN will people understand that SUSE is NOT delivering openSUSE as a unit...
Dominique
-- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org