Hijack wishes

For my birthday (coming up in April), I wish for the following in Hijack:

- Separate home/car memory for the left/right time alignment. I always want it centered at home, and only sometimes shifted in the car. If flash space is an issue, perhaps an ;@AC force_center_delay=1 option or something.

- Separate home/car memories for button brightness. Same as above if flash memory is an issue.

- Button brightness that dims with the display when the headlights get turned on.

- Remote HTTP or serial commands for voladj, and also have Charcoalgray99's XML stylesheet include these. I'd like to be able to send a command to the player for http://myempeg/voladj?low or something like that. Then the index.html homepage would have links for those commands. Surprisingly, one of the things I do often is switch the voladj level. I know I could do it by redefining the knob press and doing it remotely that way, but I already use the knob press for something else, and I often make mistakes adjusting via menus remotely.

It might also be fun to be able to send parameter sets to Voladj via http so we could play with different settings in realtime. It wouldn't have to save them, it would just be for experimentation so we can eventually stick the settings into config.ini.

You know what else is coming up in April?

Bruno

No, what?

First of the month they will announce that they've ported the Empeg Player to Windows CE.NET and that it will only play WMA files...

MY Birthday!

- Separate home/car memory for the left/right time alignment. I always want it centered at home, and only sometimes shifted in
the car. If flash space is an issue, perhaps an ;@AC force_center_delay=1 option or something.


Well, Happy Early Birthday. Have a kernel on me. (modified HJ 233, also contains my preliminary bass/treble support)

[hijack]
;@AC lr_delay = 0
;@DC lr_delay = 42

-127 <= lr_delay <= 127
delay=lr_delay*0.1ms
Value is in tenths of ms - negative values mean right channel delayed, positive = left

config.ini value will overload the value in flash upon powerup, so values that get adjusted during use also get lost upon powerdown. I couldn't see an easy way to preserve user-set values in DC mode and use config.ini values in AC mode, which would have probably been ideal for the task.

Personally I think the 8 bits of flash used for storing this delay would be better donated to saving bass and treble levels....but I'm biased.)

I'm sure that Mark will find a better way to do this by next week than I could by April

awesome! it works!

Hey genixia,

Glad you were able to pick up the ball from where I dropped it. I haven't even had time to install my empeg back in my car now that I finally have it back.

Anyways, I can't seem to get it to work. I just get this on the comport:
[hijack] ERROR: : "rl_delay = 42"

Any chance of making a patch available?

Oops, yes I definitely did try 'lr_delay' first. I just tried the other way around in case and then cut and pasted into the post.

me too!
If anyone fancies buying me a present, a tuner would be nice!

I think this is great!

With genixia and shonky prototyping new features for Hijack, a lot of issues can be worked out in advance, leading to cleaner integration into the main stream when I eventually pilfer their stuff!

But seriously, this is fantastic.

And I think we're eventually going to have to steal more flash space soon. Fortunately, I've got a much better feel for it now, and there is a fair amount of "unallocated" space in the zone we need for this.

We might just double the flash savearea completely (256 bytes instead of 128 bytes, shared with player). Gotta do it in a transparent fashion for the player, of course, hopefully such that a new player .upgrade wouldn't be too confused by it all either.

Or perhaps just have a completely separate zone for Hijack, but that might not be as safe as spewing out a single zone at emergency power down..

Cheers

Mark, is it as easy as doubling the space for the save area? I thought there was a hardware limitation where the blocks were 128 bits, and it seemed to me that if we wrote two blocks, they might not both get written before the power disappears... If that's not the case, and we can get 256 bits, that'd be wonderful, of course!

Is it time to think about saving to a small area on the disk instead?

Not quite that easy, but not too hard either.
Or at least not according to the (outdated) comments in the kernel flash memory driver (drivers/block/flash-empeg.c).

>Is it time to think about saving to a small area on the disk instead?

No, because the last thing we want to do when somebody yanks the unit out of the dash is try to write to the hard drive! Duh!

Cheers

No, because the last thing we want to do when somebody yanks the unit out of the dash is try to write to the hard drive! Duh!

Well I wasn't really thinking about the powersave stuff that you want saved every second, more for settings which you won't change much like some of the hijack configs (screen blank settings, max temp, etc.) I do agree the powersave flash block(s) would be better but I thought the disk write was another option. Not on power loss, but when you actually change a setting. I understand your point though.

Getting more flash would be great!

Sorry 'bout the problem folks, it seems as though my usually rock-solid RAID-1 /var partition got corrupted somehow, and some recent updates got lost, so I think you probably ended up DL'ing the previous kernel (bass and treble modifications only - no config.ini support for lr_delay). Hopefully that's fixed now.

patch source will be here very shortly, and the binary (based on HJ233) is here.

I've just spent an hour redefining my DNS records, as they also got screwed. Man I need to to backups.

Mark, if we got more flash we could make the config.ini/flash interactions more powerfull by reversing the order of hijack_restore_settings() and reset_hijack_options() in hijack_init().

Then we could redefine (if direction == 0) clauses in various hijack features to reload the config.ini value rather than 0.

>Tony's birthday wishlist..

Way too easy, Tony. Not that I'm complaining, since things are about to get really busy here again soon.

But I don't want to leave it until April, so here it is today. But you're not supposed to open it until your birthday, okay?

Hijack v234 is now released, will all of Tony's wishlist implemented.

Full release notes will appear shortly under a new thread.

Cheers

But you're not supposed to open it until your birthday, okay?

>Mark, if we got more flash we could make the config.ini/flash
>interactions more powerfull by reversing the order of
>hijack_restore_settings() and reset_hijack_options() in hijack_init()

That sounds like a good idea, but I'm not sure if I actually understand it.

Hijack_restore_settings has to happen as early as possible during startup, long before the drives are available (and even longer before config.ini can be read). This is because the restore_settings is responsible for telling the Empeg what kind of power it is using, AC or DC, and also because I want the menu to be running even before the player starts up, to help diagnose any troubles.

??

awwwe mark and tony are in love.. how cute

Strictly platonic, of course.

Now now, don't be jealous. Your turn (with Tony) will come someday..

extern int patch_age;   	// Age in hours.

extern char patch_status[8];

if ((patch_age < 24) && (patch_status == "Obsolete")

   printk ("The (M)Lord has spoken.");

Well, at least you didn't obsolete my bass and treble code as well, but now I'll have to port it to 234
Can I reserve 3bytes of the new flash for bass/treble levels?

Twenty-four bits sounds very extravagent. How about less, say perhaps 6 bits for each of AC and DC power modes?

?

That'll do - I can squeeze it into 6 bits/mode.

Okay, good. That still gives a slider range of about +/- 31 increments from centre, or +/- 15 increments if the bass/treble are separate sliders (I haven't pulled down the hack yet, and the last links no longer work).

EDIT: thats +/- 7 increments with separate sliders.. doh!

Cheers

Oh, and well you are at it..

Could you implement an config.ini "tuner_volume_offset" parameter?

Thanks

Or maybe "tuner_volume_boost" would be a better name.

Hmm, is my ISP playing silly buggers?...

I've doublechecked everything here, those links should be valid. (But the patch is still vs 233 and is probably broken for 234 - I'll rework it).

I have +/-12 increments per 'slider' and zero, bass and treble are separate. (I didn't do graphical sliders...(yet?!))

But still, I can encode into 6 bits. 2x25 possible states < 2^6

Hey genixia,

Still having the same problems with lr_delay not being recognised. Are you still having server problems? It seems it doesn't really matter though anymore. It looks like Mark has integrated it into the main Hijack tree now.

Yeah, for the lr_delay, you can ignore my patch..I correctly predicted that Mark would do a better job far faster

However, it should have still worked for you as I did nothing to any of the option parsing code itself. All I did was to add 'lr_delay' to the list of options being parsed, and lopan (IIRC) made it work, so...

The only thing that I can think of is that maybe you didn't put lr_delay in the [hijack] section of config.ini...?

Hmmm... It definitely was in the hijack section because hijack was wingeing about not recognising the option. Anyhoo problem is solved by other means now. Thanks, again.