Since x is viewport-relative the icon isn't necessarily placed at the physical
display boundaries so that the padding isn't always useful. In fact it does
more harm if one wants to place an icon exactly at 0 of a (non-default)
viewport.
Calling code looks still mostly fine. I've only modified list drawer to include
the padding in the call-site.
Change-Id: I6b16b3d4377c3553234667b79837adde10e0edf2
It is similar to lcd_gradient_fillrect(), except that it only draws a part
of the complete gradient. This can be used to draw only the bottom half
of a full gradient.
Change-Id: Ib47cc5237f6966e35ba07988bddbb00fd97adf96
This function supports installing a custom scroll callback. This will be
called when the scrollengine redraws the line. It allows to draw extended
styles (or anything your can possible imagine) along with the text.
It is also strictly pixel-based, the first pixel-based function that supports
scrolling.
Change-Id: I57f81ac7b3d08b877aea4cb8afa882f175ebcdfc
disconnect() needs to be called exactly once per call to init_connection().
In case of bus resets, disconnect() was not called, which led to leaking
alloc_maximum() allocated buflib handles, which led to buflib running out
of memory to allocate.
Change-Id: I03025da578dc54e48b6de6bd3e3f40feae7220a6
The old icons looked exactly like the mono version, and all >1 bpp drivers
support rendering mono bitmaps. Therefore a mono bitmap can be used which
requires less ram.
This affects only the builtin icons, not the ones used by cabbiev2.
Change-Id: I3b02b5b04fe8b4bcc69e83310871254d336b648a
Some seldomly used drawmode combinations did not work in conjunction with
alpha bitmaps and backdrops. Now all should work (see comment added) by using
more bits.
Change-Id: I2bc96ecf471fa8c1a608a321a235b9c8527b3dc5
OS X ar operates on fat libaries. In this case updating the library isn't
possible and when those change ar will only return an error. Remove the output
file prior to running ar to work around this limitation.
Change-Id: I7ebc66efd092a8e6037ae86a3658afe6b4da777f
Instead of having to update it every year just drop it. We have the build date
in the binaries we provide, and the years it has been developed can be
retrieved via git anyway.
Change-Id: Ib33ee851883146509034c405cd65552a0f67194e
This reverts commit 61a096499b.
The original issue was caused by a new structure member which caused
bmp_args::buf to be unaligned for 2-byte reads. Enforcing that alignment
should be the faster fix. Aligning to cache (while at it) should
improve bmp loading times even more.
Change-Id: I58a2caaf08c0ce46e2fb9666de628a30a36ea5f4
bpb_is_sane() used to effectively multiplying the sector size (relative
to 512 bytes) twice, which meant that filesystems with e.g. 2K sectors and
32 sectors per cluster were rejected because while this adds up to 64K
clusters (i.e. the upper limit), the calculation wrongly came to 256K.
This bug tends to affect 5.5G ipods when formatted using dosfstools.
Change-Id: Ia3f1e1303b2af953f497ccdbf23cd49c3d72e46a
On hwcodec talk.c has the entire audio buffer (not just parts of it), therefore
it must give up everything and cannot count on core_alloc_maximum() to return
the remaining space. This is equivalent to it was handled before 22e802e.
You could probaby do smarter and shrink for example the .talk clip buffer
but is it really worth it?
Change-Id: Idc3431c59fb41b05338559c615093358c5d8ed9b
