You write a simple check to catch some error condition and in the end it turns out that the error handling fails, because you can't produce the error on your system and there was a simple syntax error.
Because Slackware users wouldn't like it and it would mean replacing a thoroughly tested script with a tool, which has been tested on a very limited amount of setups so far. Don't get me wrong, I'm saying by no means that LiloSetup was bad, but it's quite young still and I'm quite sure it misses a few random special "setups" still (e.g os-prober which it relies on did until recently not detect 64bit linuxes simply because there was only a check for ld-linux.so in /lib, oh btw if somebody isn't left as clueless as me with debian bug reporting, I'm open to report it with our patch).damNageHack wrote:Why? Because of pyGTK/glade is currently used?thenktor wrote:LiloSetup cannot be used at installation time.
We could abstract the interface to add more "views" (ncurses, stdin questions list ...), just like for example with kernel sources (make config, make menuconfig, make xconfig).
And if you doubt my knowledge about writing bootloader configuration tools, I know what I'm talking about since I faced the one or other such problem with sg2conf though writing it with the great (sometimes not so great) backend provided by grub and I packaged os-prober btw reading the one or other part of it's source.
And btw another frontend increases the potential for errors in any new interface and the restructuring necessary to make the gui handling more abstract would definitely introduce the one or other behavior causing a slightly noticeable feeling of unease at the user's side.
Let's close this topic, please.
