Latest Bugs & Tasks

Syndicate content Haiku
Trac Timeline
Updated: 1 hour 22 min ago

Ticket #13043 (Cookies Not Enabled Error in Webpositive) closed

Mon, 2016-10-31 08:01
fixed:

Mh… These guys surely know how to mess up expiration dates in cookies!

Set-Cookie: TD-persist=BDC;expires=Mon, 31-Oct-2016 08:08:40 GMT;path=/;

31 oct is a sunday in this part of the world.

Set-Cookie: sitecontext=58fdf13f-cc86-4e6d-af44-3676834fdf99; Expires=Tue, 19-Oct-66 07:08:40 GMT; Path=/; Domain=.td.com; Secure
Set-Cookie: BID=dd0233c2-5fa5-473d-af30-486a7659e47d-2354188815; Expires=Sat, 21-Oct-56 07:08:40 GMT; Path=/waw/idp; Domain=td.com; Secure

Using 2-digit year is not a good idea, but is allowed by the HTTP spec. We parse those as year 56 and 66 B.C., where it should be 2056 and 2066.

Anyway, fixed in hrev50636.

Categories: Development

Ticket #13043 (Cookies Not Enabled Error in Webpositive) created

Sun, 2016-10-30 21:01

hrev50634 x86_gcc2
Going to the following URL brings up the error message:https://easyweb.td.com/waw/idp/login.htm?mode=resume
Screenshot1.png attached

Categories: Development

Ticket #13042 (Not all cookies saved when closing WebPositive) closed

Sun, 2016-10-30 19:26
fixed:

Patch applied in hrev50634, along with more fixes to cookie date parsing.

Thanks for investigating and fixing this! It will make web browsing on Haiku a little less annoying :)

Categories: Development

Ticket #12977 (Radeon HD7570 only VESA) closed

Sun, 2016-10-30 15:22
fixed
Categories: Development

Ticket #13042 (Not all cookies saved when closing WebPositive) created

Sun, 2016-10-30 13:25

When you close WebPositive, it saves all cookies to a file, but sometimes some cookies are missing. This can also be seen in the "Cookie manager" as that is missing the same cookies.
To reproduce, go to tweakers.net. It should show a remark that you should accept the cookies. When you click "Ga verder", it should store a cookie indicating that you have accepted it, but when you close WebPositive and open it again, it shows the remark again.

Categories: Development

Ticket #13041 (Antialiasing of vertical text introduces white background) created

Sun, 2016-10-30 12:58

Version: hrev50629

BFont font;
font.SetRotation(90.0);
SetFont(&font, B_FONT_ROTATION);
DrawString(strScore.ToString(), pos);


Categories: Development

Ticket #13040 (Launch scripts in ~/config/data/user_launch don't work) created

Sun, 2016-10-30 12:53

This is hrev50627.

I have a package that installs this launch script in $data/user_launch:

target desktop {
	service x-vnd.Clipdinger {
		if setting ~/config/settings/Clipdinger/Clipdinger_settings autostart
		no_safemode
		legacy
	}
}

When the package is installed in the system hierarchy, Clipdinger is started on boot. If it's installed in the home hierarchy, it's not. As the launch_daemon is supposed to launch via app signature, it shouldn't matter in which hierarchy the binary ended up in, right? Am I missing something?

Categories: Development

Ticket #13039 (Appearance: combobox for anti-aliasing type moves around) created

Sun, 2016-10-30 12:45

Version: hrev50629

Preferences / Appearance / Antialiasing

  • Select "Revert"
  • Set "Glyph hinting" to "off"
  • Set "Antialiasing type" to "Grayscale"

Now the combobox for "Antialiasing type" moves a few pixel to the left.
The position can be reset by clicking "Revert".


Categories: Development

Ticket #12799 (Unit test BClipboard::Clear1 crashes) closed

Sun, 2016-10-30 11:04
fixed:

Applied in hrev50632.

Categories: Development

Ticket #13038 (Bad shift operation in app_server overlay colorspace) created

Sun, 2016-10-30 05:33

​http://cgit.haiku-os.org/haiku/tree/src/servers/app/drawing/Overlay.cpp#n254

large_shift: In expression colorKey.alpha >> alphaShift, right shifting colorKey.alpha by more than 7 bits always yields zero. The shift amount, alphaShift, is 8.

CID 1273452

Categories: Development

Ticket #13037 (Review B_32_BIT_CONTIGUOUS memory) created

Sat, 2016-10-29 23:18

I see a few references in the code that B_32_BIT_CONTIGUOUS was a hack. Quite a few drivers use it and the requirements to leverage it aren't well documented... hell there is only one ticket ever opened about it.

kallisti5@eris haiku :) $ grep -R B_32_BIT_CONTIGUOUS
headers/os/kernel/OS.h:#define	B_32_BIT_CONTIGUOUS		6	/* B_CONTIGUOUS, < 4 GB physical address */
src/add-ons/kernel/bus_managers/firewire/util.c:		B_32_BIT_CONTIGUOUS, protection);
src/add-ons/kernel/bus_managers/scsi/emulation.cpp:		total_size, B_32_BIT_CONTIGUOUS, 0, 0, 0, &virtualRestrictions,
src/add-ons/kernel/bus_managers/usb/PhysicalMemoryAllocator.cpp:		roundedSize, B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/bus_managers/usb/Stack.cpp:		B_32_BIT_CONTIGUOUS, 0);
src/add-ons/kernel/busses/ata/silicon_image_3112/silicon_image_3112.c:		B_ANY_KERNEL_ADDRESS, prdtSize, B_32_BIT_CONTIGUOUS, 0);
src/add-ons/kernel/busses/ide/silicon_image_3112/silicon_image_3112.c:		B_ANY_KERNEL_ADDRESS, prdtSize, B_32_BIT_CONTIGUOUS, 0);
src/add-ons/kernel/busses/scsi/53c8xx/53c8xx.c:			B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/busses/scsi/buslogic/buslogic.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/audio/ac97/auich/util.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/audio/ac97/auvia/util.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/audio/ac97/es1370/util.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/audio/ac97/geode/geode_controller.cpp:		B_ANY_KERNEL_ADDRESS, alloc, B_32_BIT_CONTIGUOUS,
src/add-ons/kernel/drivers/audio/ac97/geode/geode_controller.cpp:		B_32_BIT_CONTIGUOUS, 0);
src/add-ons/kernel/drivers/audio/ac97/ich/util.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/audio/ac97/ichaudio/lala/util.c:		B_32_BIT_CONTIGUOUS, protection);
src/add-ons/kernel/drivers/audio/ac97/sis7018/Stream.cpp:				B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/audio/echo/util.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/audio/emuxki/util.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/dvb/cx23882/util.c:		B_32_BIT_CONTIGUOUS, protection);
src/add-ons/kernel/drivers/graphics/nvidia/driver.c:			B_32_BIT_CONTIGUOUS, /* GPU always needs access */
src/add-ons/kernel/drivers/graphics/nvidia_gpgpu/driver.c:			B_32_BIT_CONTIGUOUS, /* GPU always needs access */
src/add-ons/kernel/drivers/graphics/radeon/PCI_GART.c:		B_32_BIT_CONTIGUOUS,
src/add-ons/kernel/drivers/network/dp83815/driver.c:		B_ANY_KERNEL_ADDRESS, pages * B_PAGE_SIZE, B_32_BIT_CONTIGUOUS,
src/add-ons/kernel/drivers/network/dp83815/util.c:		B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/drivers/network/rtl8169/util.c:		B_32_BIT_CONTIGUOUS, protection);
src/add-ons/kernel/drivers/network/wlan/ipw2100/ipw2100.cpp:		size, B_32_BIT_CONTIGUOUS, B_READ_AREA | B_WRITE_AREA);
src/add-ons/kernel/generic/ata_adapter/ata_adapter.cpp:		prdt_size, B_32_BIT_CONTIGUOUS, 0);
src/add-ons/kernel/generic/ide_adapter/ide_adapter.cpp:		prdt_size, B_32_BIT_CONTIGUOUS, 0);
src/kits/debugger/user_interface/util/UiUtils.cpp:		case B_32_BIT_CONTIGUOUS:
src/system/kernel/vm/vm.cpp:		case B_32_BIT_CONTIGUOUS:
src/system/kernel/vm/vm.cpp:			lock = B_32_BIT_CONTIGUOUS;

usb stack:

area_id
Stack::AllocateArea(void **logicalAddress, phys_addr_t *physicalAddress, size_t size,
    const char *name)
{
    TRACE("allocating %ld bytes for %s\n", size, name);
    void *logAddress;
    size = (size + B_PAGE_SIZE - 1) & ~(B_PAGE_SIZE - 1);
    area_id area = create_area(name, &logAddress, B_ANY_KERNEL_ADDRESS, size,
        B_32_BIT_CONTIGUOUS, 0);
        // TODO: Use B_CONTIGUOUS when the TODOs regarding 64 bit physical
        // addresses are fixed (if possible).
Categories: Development

Ticket #13036 (cuserid() - behaviour different to BeOS) created

Sat, 2016-10-29 16:37

Version: hrev50595
Reference: man cuserid

char[1024] strTemp;
const char* strUsername = cuserid(strTemp);

Under R5, the function would return the current username, under Haiku it returns an invalid pointer.

Note: this can be closed immediately, but the difference in behaviour should be documented.

Categories: Development

Ticket #13035 (Stack&Tile: Tabs don't redraw after right-click to send-to-back) created

Sat, 2016-10-29 12:28

Version: hrev50595

Steps to reproduce:

  • create 4 new directories
  • open them
  • stack two of them together
  • select the second tab of the stack
  • right-click on the tab to send it to the back
  • The other stack should get the focus, one of its tabs becomes yellow, but the old stack still has a yellow tag


Video: ​https://herzig-net.de/prog/haiku/trac/Haiku-2016-10-29T12-14-21-715230000Z.webm

Categories: Development

Ticket #12094 (BCatalogData has no way of querying which keys are available) closed

Sat, 2016-10-29 09:18
no change required:

You can use HashMapCatalog::GetWalker to get a CatalogWalker instance, which allows to do that.

Categories: Development

Ticket #13034 (show bitmap icons in HaikuDepot app) closed

Fri, 2016-10-28 15:04
duplicate:

I agree with you.

Categories: Development

Ticket #13034 (show bitmap icons in HaikuDepot app) created

Fri, 2016-10-28 12:28

The Haikdepot web app allows one to select either HVIF or PNG files as icons for packages. However, in the desktop app, only HVIF icons are displayed.In the attached screenshots, see the status of the apps 256 and 4colors_j, for example.

My repo has over 900 packages, around half of which are old BeOS apps. Their icons are part of their brand and history, and it would not respect the original authors IMHO to just create all-new icons for them. My own lack of graphics skill does not make it feasible to re-create these icons as vector icons from scratch.

I request that PNG icons also be displayed in the Haikudepot desktop app.

Categories: Development

Ticket #13033 (DriveSetup - Add Disk Wipe (write zeros to drive)) created

Thu, 2016-10-27 16:00

Extend the capabilities of the Haiku Installer to be able to wipe a hard disk by writing zeros to drive in preparation for partitioning and formatting.

Categories: Development

Ticket #13032 (SerialConnect logs not visible with StyledEdit; error mesaage with Pe) created

Wed, 2016-10-26 23:54

hrev50621 x86_gcc2
SerialConnect->File->Log to file...
The logs created by the app are blank when opened in StyledEdit
Opening them with Pe brings up an error message
See attached screenshot

Categories: Development