Latest Bugs & Tasks

Should be fixed in hrev45321. diver can you please verify and reopen if not fixed?

There doesn't seem to be a good way of seeing what dB level each fader is set to in the audio mixer. When dragging a fader (which risks moving it unintentionally) it shows a number in a popup below the fader, but this number has extra digits at the end, and no "dB" on the end. It is also cut off width-wise and sits immediately below the fader, where it is mostly covered by the mouse pointer.

Would it be possible to put an indicator of the current setting at the bottom, below where it says "To output", "To master" etc.? Or alternatively use a tooltip that pops up when you mouse over the fader knob, but does not require clicking the fader.

Graticule markings like on a real mixer next to each fader would help too, but there needs to be some way of seeing the exact level.

It is also quite difficult to set unity gain (0dB) - something like double-clicking, shift/option/cmd+clicking would work. I prefer the latter, a double-click seems like it might be quite dangerous. The level could suddenly jump up if you've got a finnicky mouse or touchpad. Furthermore, using yet another key could "gear it down" to give finer adjustment possibility.
Some other systems use a "centre-detent." Please don't do that, they're awful! They're particularly bad if you're actually trying to mix audio.

Finally, some way of metering the overall output level would be useful. Even a crude inaccurate "clip" indicator that illuminates when a sample codeword exceeding +/-1 has been detected in the last second would be useful. I realise this might be harder due to the split GUI / backend nature of the mixer, so consider this aspect lower priority.
However I believe even for consumer audio a clip indicator would be useful. People need to be trained in setting levels properly. I swear, if I hear one more Youtube video that's slamming into hard clipping...

If a full meter is going to be implemented, I'm able to help by writing the DSP code. True peak (reconstruction filtered) and VU-characteristic RMS or average can be converted into number of pixels tall to fill the meter. I have no clue architecturally how to get the peak data from the mixer back end to the GUI however.

It took me a while to figure out how the "input channel remapping" and "output channel remapping" works in the audio mixer. I got confused by the fact the mixdown busses are given names like "left, centre, right" etc.

IMHO these changes would make it more logical to someone coming from an audio background (i.e. anyone who is likely to enable "enable input/output channel remapping" in setup) and would better support soundcards with lots of I/O.
32-input/32-output soundcards are available and Haiku already has support for a 16-input/16-output soundcard (Echo Layla 3G) which cannot be fully utilised with the current number of busses (12: left,right,centre,sub,rear-left,rear-right,front-left-centre,front-right-centre,back-centre,side-left,side-right,mono)

In summary the changes I propose are:

• Increase number of mixing busses to 16 or 32.

• Rename the mix busses so they're numbered 1-32, with the name down-played in brackets after the number to show the user this is the recommended use of the buss in a surround-sound environment, but not hard-wired.

• Rename the input maping tab "channel routings" or similar. This is more in keeping with the naming used on a real audio mixer.

• Rename the output mapping tab "buss routings", "output routings", or similar.

• Replace the "Allow output channel remapping" and "Allow input channel remapping" checkboxes with a single checkbox that enables both, called something like "show routing tabs." Or just get rid of it entirely, if the user doesn't understand those tabs, they can just ignore them.

• For some reason after clicking each checkbox in the "input remapping" tab, it defaults back to the mixer. Grrrrr!! If it's too difficult to change, consider moving the channel routing checkboxes into the mixer pane along with the faders, like a real analogue mixer has. This may need an option to show/hide them due to the width consumed (especially on a program that outputs many channels)

• If possible, offer the user some way to load and save mixer settings (as a file) so different routings (and mixer levels) can be saved and recalled. If this is too hard, offer a choice of two hard-wired routings in the setup tab which can be recalled: surround sound (which works like the current Haiku mixer, sending the "mono" buss to left and right, etc.) and individual channels (which routes the busses 1:1 at unity gain to the soundcard's outputs, one output per buss, one buss per output, if there are less than 32 soundcard outputs the higher numbered busses are not routed anywhere)

• Above all else, reference how real audio mixers behave when doing anything with Haiku's mixer. Think outside the box of surround sound! I am happy to assist with designing features or changes, but my BeAPI coding experience is limited.

Fixed in hrev45317 but my above guess was wrong, Deskbar got the message fine. The problem was that views are considered hidden if their parent window is hidden and I didn't properly handle that case.

Keyboard randomly stops functioning. Have to unplug and replug to regain functionality.

Logs show no errors when keyboard goes dead... usb_hid error is seen on plugging in the keyboard however...

KERN: usb hub 38: port 5: new device connected
KERN: usb hub 100: port 1: new device connected
KERN: usb_hid: no handlers for hid device
KERN: usb_hid: keyboard device unhandled control 0x00002710

Keyboard USB info

05ac:0220 /dev/bus/usb/4/1/5/1 "Apple, Inc." "Aluminum Keyboard (ANSI)" ver. 0069

~> listusb -v /dev/bus/usb/4/1/5/1
[Device /dev/bus/usb/4/1/5/1]
    Class .................. 0x00 (Per-interface classes)
    Subclass ............... 0x00
    Protocol ............... 0x00
    Max Endpoint 0 Packet .. 8
    USB Version ............ 0x0200
    Vendor ID .............. 0x05ac (Apple, Inc.)
    Product ID ............. 0x0220 (Aluminum Keyboard (ANSI))
    Product Version ........ 0x0069
    Manufacturer String .... "Apple, Inc"
    Product String ......... "Apple Keyboard"
    Serial Number .......... ""
    [Configuration 0]
        Configuration String . ""
        [Interface 0]
            [Alternate 0 active]
                Class .............. 0x03 (HID)
                Subclass ........... 0x01
                Protocol ........... 0x01
                Interface String ... ""
                [Endpoint 0]
                    MaxPacketSize .... 8
                    Interval ......... 10
                    Type ............. Interrupt
                    Direction ........ Input
                [Descriptor 0]
                    Type ............. 0x21
                    Data ............. 11 01 21 01 22 4b 00
        [Interface 1]
            [Alternate 0 active]
                Class .............. 0x03 (HID)
                Subclass ........... 0x00
                Protocol ........... 0x00
                Interface String ... ""
                [Endpoint 0]
                    MaxPacketSize .... 1
                    Interval ......... 10
                    Type ............. Interrupt
                    Direction ........ Input
                [Descriptor 0]
                    Type ............. 0x21
                    Data ............. 11 01 00 01 22 2f 00

Keyboard hub info:

05ac:1006 /dev/bus/usb/4/1/5/hub "Apple, Inc." "Hub in Aluminum Keyboard" ver. 9615

~> listusb -v /dev/bus/usb/4/1/5/hub
[Device /dev/bus/usb/4/1/5/hub]
    Class .................. 0x09 (Hub)
    Subclass ............... 0x00
    Protocol ............... 0x01
    Max Endpoint 0 Packet .. 64
    USB Version ............ 0x0200
    Vendor ID .............. 0x05ac (Apple, Inc.)
    Product ID ............. 0x1006 (Hub in Aluminum Keyboard)
    Product Version ........ 0x9615
    Manufacturer String .... ""
    Product String ......... ""
    Serial Number .......... ""
    [Configuration 0]
        Configuration String . ""
        [Interface 0]
            [Alternate 0 active]
                Class .............. 0x09 (Hub)
                Subclass ........... 0x00
                Protocol ........... 0x00
                Interface String ... ""
                [Endpoint 0]
                    MaxPacketSize .... 1
                    Interval ......... 12
                    Type ............. Interrupt
                    Direction ........ Input
    Hub ports count......... 3
    Hub Controller Current.. 200mA
      Port 1 status....... 0100.0000 Power
      Port 2 status....... 0303.0000 Connect Enable Power Lowspeed
      Port 3 status....... 0100.0000 Power

My Centrino Notebook Sony Z1 XMP does not boot with haiku-nightly-hrev45316-x86gcc2hybrid-cd.
It went to KDL. Screenshot enclosed. This notebook is fully supported by Haiku.
The last working revision i tested was haiku-nightly-hrev45234-x86gcc2hybrid-cd.
Cpuidle was enabled between these revisions and cpuidle is mentioned in the KDL so maybe this could be the problem.

hrev45285 gcc2h.

To reproduce:

• Launch Mail.
• Edit -> Settings.
• Select "Button bar: Hide"
• "OK"
• Exit Mail. Launch Mail.
• The button bar is not visible.
• Go into settings and enable either "Show icons & labels" or "Show icons only".
• Crash. debug information is attached.

Both GLTeapot and Haiku3d crash with segment violation on _mesa_resize_framebuffer. I am using an AMD display controller reported as "Turks [Radeon HD 6570]", debug reports attached.

Problem: After Haiku finishes booting (just after the rocket ship BLASTS OFF!) the screen becomes black, and the monitor stops receiving input and goes into standby mode.

When: This problem started after hrev45219 "radeon_hd: Fix DisplayPort link training" (​http://cgit.haiku-os.org/haiku/commit/?id=00bc40ad588f4bd59cd7574788b5dc111b62198a)

Hardware: Radeon HD 5450 (More details: ​https://wiki.archlinux.org/index.php/User:drcouzelis)

Please let me know if there is more information I can provide or are any tests you'd like me to run. Thank you!

I make icons from PNG to SVG and then to SVG to HVIF and i always use autotrace with the same parameters. But sometimes Icon-O-Matic rejects to export as HVIF.
Screenshot attached.

When I mount my Linux partition (a ~100GB btrfs volume), the volume icon shows up on the desktop without any errors, but when I open it, I just get a folder showing the Disks icon, which leads me back to a list of the mounted volumes. It never opens the actual btrfs volume.

This is hrev45308

There is no message in the syslog when mounting the volume, but the following message shows up each time I try to actually open the mounted volume.

2013-02-21 14:26:27 KERN: btrfs: Inode::UpdateNodeFromDisk(): Couldn't find inode 259
2013-02-21 14:26:27 KERN: btrfs: get_vnode: InitCheck() failed. Error: No such file or directory

This is hrev45313.

Open Deskbar preferences...-> Window -> enable Auto-hide
Right click clock in Deskbar shelf->Time preferences

Show clock in Deskbar is not enabled and resets its' checkbox when I try to check it.

You are expected to allocate a dirent structure big enough for the names of the files in the directory. The d_name array in the struct definition only has space for one character. In this case, readdir_r scribbles over plugindir, causing the second call to readdir_r to fail and the call to closedir to crash.

The reason it works on Linux is because struct dirent has a larger d_name. But since POSIX doesn't specify the length of the array, portable code shouldn't rely on that.

I have noticed that inside "FileType" window, Haiku icon format has generic infos (look at the screenshot), and is under the wrong category: is under "application" and should be located inside "image".

hrev45280

p.s: i have insert this ticket under Applications > Tracker since seems that is not present a category related to file types.

Since hrev45280 i have found a strange behaviour about focus.
The first thing is: if in an application a tooltip appears, this tooltip will steal focus to application and you have to click with mouse in app window to take again the focus.
second: if inside /system folder you attempt to rename a file, obviously as usual, an alert prompt you about. You click on rename and then the Tracker window will lost the focus and made impossible to rename the file, because the text field for renaming file is present only in focused windows. To be more clear i made a little video (attached) of this.

I am using Haiku to build the following program

#include <stdlib.h>
#include <stdio.h>

#include <errno.h>
#include <dirent.h>

int main(void) {
  DIR *plugindir = opendir("loaders");
  struct dirent direntry;
  struct dirent *dirresult;
  int i;
  if (plugindir == NULL) {
    return;
  }
  do {
#ifdef CRASH
    int res = readdir_r(plugindir, &direntry, &dirresult);
    if (res != 0) {
     printf("fail %s\n",strerror(res));
      break;
    }
#else
  dirresult = readdir(plugindir);
#endif

    if (dirresult == NULL)
      break;
    printf("file %s\n",dirresult->d_name);
  } while (1);
  closedir(plugindir);
  return 0;
}

I use native gcc 2.95.3 (that is, no cross-compiling, gcc runs on Haiku itself) to build it (by the way, are newer version of gcc available?). If I compile it with -DCRASH, this is the output.

file .
fail Operation not supported
cr: /storage/haiku-releases/r1alpha4/haiku/src/system/libroot/posix/malloc/processheap.cpp:166:void BPrivate::processHeap::free(void *): sb
Kill Thread

After installing alpha 4.1 over alpha 2 on my iMac, the mouse would freeze when booting, even in safe mode. It would freeze at the point where the Tracker or Deskbar is launched. If I commented out the lines from Bootscript launching the tracker and deskbar, and launched terminal instead, I could look around using the terminal. I finally got it to boot, as follows, but I don't know if it is reliably repeatable.

1. boot from the installer CD, holding shift key. When the safe mode window comes up, choose "select safe mode options". Of these choose the 1st (safe mode) and the 4th (fail safe video). Then boot into the desktop from the CD.

2. the problem seems to be with virtual memory. Whenever it won't boot from the hard drive, and I reboot from the CD, virtual memory is always maxed out, asking for a humungous swap file. Open the virtual memory preferences, and turn off "choose swap size automatically". Move the slider to select a small swap file size.

3. Open hard disk/system/boot/Bootscript using Pe. Comment out every line that says

if (SafeMode != 'yes') then
and comment out the matching fi.
Do the same for hard disk/system/boot/SetupEnvironment.
This is in case you need to reboot using safe mode.

4. Reboot, this time from the hard disk. Hold shift key. When the safe mode window comes up, choose "select safe mode options". Then only choose "fail safe video" out of the options. Continue booting. If it doesn't boot, check the virtual memory prefs, and try again. Sometimes the second attempt will boot, even if the first one doesn't. If it still doesn't work, try again, this time choosing "safe mode" and "fail safe video" out of the list of options. I can't get it to boot without choosing "fail safe video", even tho Haiku understands the video card perfectly well.

5. After getting it to work, I made the mistake of shutting down and powering off. The next day it wouldn't reboot. Eventually I got it working again, using the above method.

When connecting to an encrypted wireless network, a window opens asking for a password.

If this window is double-clicked to hide it, it disappears without a trace - it is not listed in the Deskbar window list so it cannot be retrieved.

Clicking the wireless network again in the Deskbar network selector does not bring the window back. I can't seem to find any way to bring the window back again so the password can be entered, besides rebooting the system.

Apologies if this bug has already been filed, but searching for combinations of hiding+network seems to be returning stuff about hidden networks, nothing about hiding windows.

Neighbour is polishing the barrel of his shotgun, I'll keep this quick..

The automatic connection to the first open network Haiku sees at boot up is potentially very dangerous. My network is WPA. It is really easy to forget to select the correct network, and end up walking right into sticky situations:

• Checking e-mail: plaintext password has been splattered all over the RF spectrum.
• Checking FurAffinity, even more stuff splattered all over, and if the neighbour has logging enabled on their router I'm done for.
• Downloading ISOs, two and counting.
• Trying to FTP into my private server. Of course it fails, but if my neighbour happened to be watching, they could theoretically delopy an FTP server and trap me, taking my password as I attempt to log into it.
• Luckily the neighbour's network has not been used for private IRC chat yet. If that sort of stuff gets leaked, I'm just going to dig a hole and bury myself.
• May end up in prison or at least fined, connecting to a neighbour's network is technically illegal, especially if you steal gigabytes of their bandwidth.

Filed as an enhancement rather than a bug because there seems to be a workaround by using the commandline to connect, and putting it in the user boot script (I haven't tested this yet)

I do hope this behaviour will be changed before R1, I can't really see the utility of connecting automatically to some random network that you have no idea about, but clearly there is potential for harm by connecting.