[syslinux] Menu system bug - MENU DEFAULT not working

[Ady]

Subject:        	Re: [syslinux] Menu system bug - MENU DEFAULT not 
working
Date sent:      	Thu, 11 Oct 2012 21:49:09 +0100

> On Thu, 2012-10-11 at 20:04 +0200, Ady wrote:
> > I kept thinking about SUBmenus, even after Matt's comment that his 
> > patch would affect not just MENU GOTO but all MENU DEFAULT behavior 
> > regarding SUBmenus.
> > 
> > But then I started to think about the PARENT menu behavior. When 
> > going back from a submenu to its parent menu using [ESC], the current 
> > behavior (as of 4.06pre12) is already expected by users. To be 
> > consistent, the behavior going up to the parent menu shouldn't be 
> > changed.
> 
> Exactly. You've hit the nail on the head. People will rely on this
> behaviour. We can't regress.
> 
> > I don't have the knowledge to test the patch alone, so I'll have to 
> > wait for another prerelease so to confirm if this is the case or not.
> 
> -pre13 will be out in the next hour or so.
> 
> -- 
> Matt Fleming, Intel Open Source Technology Center
> 
> 

Hello Matt,

I tested 4.06pre13 MENU DEFAULT behavior in relation to MENU GOTO and 
multiple levels of submenus navigation. I used menu.c32 (not 
vesamenu.c32), ldlinux.sys (both with no modification at all) and 
with different cfg files, including the one from Tomas.

Regarding submenus, I used MENU BEGIN + MENU END pairs, with and 
without tagnames. I have NOT tested "[MENU] INCLUDE tagname" YET. I 
haven't tested "MENU EXIT tagname" either.

Observations:
1_ The navigation through multiple levels of parent-child (sub)menus 
is consistent with previous versions.
2_ Using Tomas' cfg file (and some variations too), MENU GOTO seems 
to "almost" work correctly and as expected.
3_ I don't know if this is important or not but... I am able to 
"break" the MENU GOTO + MENU DEFAULT, but the specific pattern is not 
so clear yet. Here is how I "broke" it.

I boot ldlinux.sys from 4.06pre13. I use the submenus from Tomas, but 
I use menu.c32, not vesamenu.c32 (I deleted all that is not relevant 
to the test with menu.c32, and I haven't even tested actually booting 
any OS kernel from the menu).

Now, while I am over some entry in the middle (one before the last 
one, or two before the last one), I press the space bar and the 
selection goes down the list of entries. Then I may press [ENTER], or 
the arrow keys, or the space bar again ("playing" with the menu and 
watching the results). At some point, while pressing [ENTER] again 
and again and again, the selection "jumps" from one entry line to 
another (which shouldn't happen in this particular cfg file). The 
jump is not seen after _every_ [ENTER], but it repeats itself in some 
way or another (keep "playing" and it shows up again).

I can replicate this strange behavior, but I am still not clear about 
the exact pattern and steps.

This is probably an edge case, but maybe someone else noticed it too.

BTW, I don't know which exactly is the expected result from pressing 
the space bar in menu.c32 from 4.06pre13, so maybe this is just a 
"stup1d-proof" test.

Regards,
Ady.