Okay, the New Feature Of The Day is a minor one: I've added "nocheck" to the "SITE RW" command in Hijack, so that remounts complete instantly rather than taking upwards of 30 seconds or so. This is the same way that Emplode does it, and the way I ought to have done it originally.

Cheers

Hey Mark, what's in the kitchen for future Hijack releases? Are you now happy with where kftpd and khttpd are?

Here are some things that either I've seen suggested on the BBS or have personally been lusting after...

• User configurable Hijack menu. Remove/re-order the various "stock" menu items.
• n-level menus. Like maybe a "Debug" submenu could contain "show flash savearea", "button codes display", etc.. Maybe allow userland programs to create a submenu with their own options... Lot of possibilities.
• A mmap buffer for the display writes instead of the ioctl() interface
• Alarm clock
  

Any of these excite you? Or do you plan on just tweaking existing functionality now? The one I'm most interested in for my own use is the display buffer, but I know a lot of people wanted the alarm clock. I think the configurable user menus and the ability to cascade submenus would be a nice addition as well.

Anyway I just wanted to check in and see where your roadmap is going, if you have one.

The "remove menu item" option is already incorporated in Hijack, as in

menu_remove=Filesystem Check on Sync

n-level menus won't happen for a while, if ever.

alarm clock is a perfect example of a userland application that needs to be written.

The Hijack website has (somewhere) a feature list of present and planned stuff, which is basically my TO-DO list. Something missing from it is I hope to make the quick-knob-press-PopUp menu into a configurable item, perhaps with "tree" capability. It looks easy now that the IR-translate got cleaned up a bit, so I might add that to the feature list sooner rather than later.

But right now, I'm waiting for the next beta, to see what it breaks, and then the Sony Stalk interface. The Stalk stuff is interesting because it gives us access to at least one new player command: "previous visual", which is not on the "stock remote" command set.

Cheers

alarm clock is a perfect example of a userland application that needs to be written.

Hmm. I guess what might be a little confusing is that there's already a countdown timer and I thought the next logical step was going to be an alarm.

Maybe I'll give it a whirl. A beeping alarm is easy, something I could probably do in an hour. Playing an MP3 will require some work because there's no easy way to select specific MP3's in the player, unless you count sending IR codes like "down, down, right, right, right, right, down, right, down" to get to a certain song/playlist. Unless I just had it hit "play" and play whatever song is currently paused... hmm. Maybe that's a good first cut, anyway.

Maybe I'll throw in a configurable beeping alarm like the old school SysOp chat pages in various BBS softwares. That'll take me back some.

Like I said, I don't care much for the alarm since I have an alarm clock that wakes me up just fine, but I've seen enough requests that it's probably worthwhile.

I'm not if this has been fixed or not. But I'm using v141 of HiJack. And I have "Restore DC/Car Visuals" enabled, and it seems to be busted! The visuals are not restored from a power down.

Just thought I'd mention it.

And I have "Restore DC/Car Visuals" enabled, and it seems to be busted! The visuals are not restored from a power down.

I was unable to get this feature working, as well, and then it magically started working for me yesterday. I do not know why.

Well, please retest with v143 or later, and if still broken, GIVE ME ENOUGH INFORMATION TO HELP YOU, damnit!

It works fine for me, but maybe you have discovered a combination of INFO and VISUALS for which it mucks up. So let me know exactly what state it was in at powerdown, so that I can try the same thing here. Also, I need to know what type of player hardware you have (mk1, mk2, or mk2a) just in case it makes a difference in the solution.

One thing I just noticed while retesting this feature, is that mk2 and mk2a units behave differently -- one will come up in standby (because the "ignition" sense is not active), and the other defaults to non-standby (because ignition sense floats the other way..).

-ml

Okay, Mark, me and my kooky IR translations again... I'm trying to have my Shift button function as a shift button only when the UI isn't active (there are reasons, I promise.) In other situations (when there's a UI active) I'd like a press of that button to send "cancel" and a long press to send "OK." So I tried this:

[ir_translate]

; The shift key

AD520C.LU=20df12                ; Long "ATT" = OK when UI active

AD520C.U=20df0c         ; "ATT" = Cancel when UI active

AD520C.LN=20df03.L,FFFFFFFF              ; press and hold "ATT" = standby

AD520C.N=FFFFFFFF.S             ; "ATT" = Shift Key when UI not active

Unfortunately for some reason my shift key no longer works when I do this. Any reason you can think of offhand? Besides the fact that this is a crazy combination of options...

Your kooky translations are <em>easy</em> this time.

I know, because something is broken that you don't know about yet.

Basically, ".U" and ".N" ALWAYS match on the left of the equal sign. This is a limitation in how the IR translate chain is implemented, and is not easy to get around. I was hoping you wouldn't notice until after I do the Stalk coding, which will probably involve fixing the IR chain..

Anyway, the point is, the first .U or .N translation that matches a given button will ALWAYS apply (unless other qualifiers are given). But to get around it, you can do something like:

somebutton=xxxxxx.U,yyyyyyy.N,zzzzzzz.U,wwwwwww.N,kkkkkkk,oooooo.N

Cheers

-ml

Okay, while I'm chewing on this I found something else.. I have a /usr partition on /dev/hda2. I'd like my makefiles to, if something builds properly, FTP it to my Empeg. So I try to do a SITE RW but it doesn't mount my /usr partition read-write. Any way you can check for the existence of another partition and mount that RW? Or allow a special RW command to mount a user-supplied mount point RW?

Er, wait, I thought commas separated each code in a multi-code sequence... From your syntax it looks like you're using them to separate different codes for a single button in various situations... I'm confused..

I tried this:

AD520C=20df0c.U,FFFFFFFF.NS

AD520C.L=20df12.U,20df03.NL

Which is what I think you were suggesting... And my shift key doesn't shift.

Should I just wait until you debug the IR trans chain?

No, the idea is that when you use .N and .U on the right of the equal sign, they just "qualify" the individual button codes as they are sent through the chain.. the ones that don't match are discarded from the translation on-the-fly, and the rest of the sequence continues..

But the check for menus is only done once, at the start of the chain.

Got it? It works!

>I tried this:
>
> AD520C=20df0c.U,FFFFFFFF.NS
> AD520C.L=20df12.U,20df03.NL

I'm not sure what that was supposed to do, but it says "when I press AD520C, send 20df0c if menus are active, otherwise toggle the SHIFT setting. The second line never gets looked at, since the first line always matches that button.

Cheers

Mmm.. yes.. I suppose I could have the in-kernel Hijack "remount" take a walk through /proc/mounts and just do all of the disk-based filesystems it finds therein.. sound reasonable?

-ml

Okay well the first line isn't working. No menus were active, yet the shift wasn't being toggled..

So how would I move the 2nd line into the first? Since .L means different things when on the left or the right... On the left it means "a long press sends..." where as on the right it means "send a long press of this button..."

Mmm.. yes.. I suppose I could have the in-kernel Hijack "remount" take a walk through /proc/mounts and just do all of the disk-based filesystems it finds therein.. sound reasonable?

I like it.

If you want the second line to work, just move it to be above the first line, so that .L gets looked for before the default match.

For your other problem, '.S' on a null (ffffff) entry doesn't work.. but I'll put it in v146 for you, as of... now!

Also new in v146 is SITE RO/RW now loop over all ext2 filesystems listed in /proc/mounts

Cheers

Hey Mark I can't use the v146 patch because it's apparently diffed against that secret beta we're not supposed to know about.

This can be remedied by:
1) Patching against 2.0b7
2) Sending me the new beta.

I prefer option #2.

Huh?

The binaries I supply work fine with any v2.00 beta, old, new, or secret.

Are you trying to use the source patch? I thought maybe that should work, too, but I'll double-check it now.

Okay well I can't seem to patch v146 properly but v143 completely ignores ir_translates with .U or .N on the right side... Like:

AD520C.L=20df12.U,20df03.NL
AD520C=20df0c.U,123.NS

Gets ignored, (the IRtrans debug doesn't show them) but

AD520C.L=20df03.L
AD520C=123.S

Seems to be okay.

I always use the source patch because I have a couple of my own things in the kernel.

This is sad. That's three releases since I got home from work. I haven't even had dinner yet.

There's a gag about "three impossible software releases before breakfast" in there somewhere, but I'm going to go eat rather than refine it.

Okay, grab the patch again and try it now..

Hrm everything except empeg_rds.c patches... Will I be okay to run it without it?

patching file arch/arm/special/empeg_rds.c
Hunk #1 FAILED at 677.
Hunk #2 FAILED at 736.
2 out of 2 hunks FAILED -- saving rejects to file arch/arm/special/empeg_rds.c.rej

You must be missing the RDS patch, that hijack sits on top of (??). Anyway, you can skip the .rej for it.

-ml

I got v146 on now, it was the RDS patch. I'm still not sure if I'm Doing The Right Thing, but it's still skipping over my ir_translate entries, it must not like them...

AD520C.L=20df12.U,20df03.NL

AD520C=20df0c.U,123.NS

The debug output shows no translations for AD520C.

Ugh.. I confess.. I was being distracted by certain company at the time I put .U/.N into the kernel, and it's been nagging at me ever since.. like I forgot do add code somewhere.. and .. well... differential patch attached.

Cheers

Hey, you don't need to explain or apologize, this is an iterative process and your tireless efforts are much appreciated here.

Anyway, I'm not sure what's going on now, it seems that now the AD520C button correctly shifts when the UI isn't on, and correctly cancels when the UI *is* on, but when it cancels, it also changes the shift lock status. That leads me to believe that when two translations are on the right side, as in:

AD520C=20df0c.U,123.NS

It's sending them both when it's in the UI mode...

Anyway I'm going to put this effort down for awhile and work on other stuff... I'll let you do the same... If you come back to the IR translations at some point, let me know and I'll try to get this working.

Puzzling - hijack site has link to v147 but it doesn't appear to exist. v146 exists and works fine. Ah well, if I wait a bit we'll be past 150 soon anyway.

Mmm.. processing of shift-toggle was happening too early in the chain. I've moved it around in v148, but have not tested IR trans in that release yet. Feel Free.

Also, I cannot figure out your email address from your profile..

Mmm.. processing of shift-toggle was happening too early in the chain. I've moved it around in v148, but have not tested IR trans in that release yet. Feel Free.

Will do, soon as I can get my breakfast and a cup of coffee...

Also, I cannot figure out your email address from your profile

Hmm I guess my spamproofing is also emailproofing it... There's a more legible version in my sig...