Current Server Time: 18:20 (Central Europe)

#apertus IRC Channel Logs

2020/03/15

Timezone: UTC


23:35
RexOrCine
left the channel
23:38
RexOrCine
joined the channel
01:41
rajdeep
joined the channel
01:46
rajdeep
left the channel
01:47
rajdeep
joined the channel
01:48
rajdeep
I saw metal_dent[m] has made a good progress in T1162. I think we both cannot do the select the same project/idea for GSOC. Is there a way by which both of us would be able to participate? I'm also doing T1162
01:53
Bertl
maybe check out T1163 as well?
01:55
Bertl
if you are creative, you can also suggest your own idea as T727, but it has to be a pretty good and solid suggestion
02:02
Bertl
off to bed now ... have a good one everyone!
02:03
Bertl
changed nick to: Bertl_zZ
02:05
rajdeep__
joined the channel
02:08
rajdeep
left the channel
05:14
RexOrCine
left the channel
05:15
RexOrCine
joined the channel
05:48
jason
joined the channel
05:49
jason
changed nick to: Guest11454
05:49
Guest11454
left the channel
06:43
rajdeep__
left the channel
07:05
rajdeep
joined the channel
07:09
rajdeep
I am getting a black image when I replaced the `apertus_logo` struct with a new struct called `apertus_text`(I converted one of the disected image to png and then to c source header)
07:32
rajdeep
left the channel
07:32
rajdeep
joined the channel
08:16
rajdeep
left the channel
08:23
BAndiT1983|away
changed nick to: BAndiT1983
08:24
BAndiT1983
rajdeep: if you commit your code, then i could check if something is off
09:57
Bertl_zZ
changed nick to: Bertl
09:57
Bertl
morning folks!
09:58
se6ast1an
good day
09:59
BAndiT1983
hi
10:25
HEGAZY
joined the channel
10:27
HEGAZY
left the channel
11:04
rajdeep
joined the channel
11:05
rajdeep
https://github.com/Razdeep/AXIOM-Remote/commit/5eacd37f063bbe6f63e735300c9d015ad1d465fd
11:05
rajdeep
@Bandit1983 here's what I did
11:06
BAndiT1983
are the width and height right?
11:06
BAndiT1983
what format does the data have?
11:07
BAndiT1983
are you trying to solve T1162?
11:08
BAndiT1983
by the way, just exporting from GIMP won't work usually, as the data has to be in RGB565 format, don't know if GIMP supports it directly, used online converter to create the image back then
11:10
BAndiT1983
ok, just googled and GIMP supports it, but one has to select proper options
11:10
BAndiT1983
without RLE compression of course, it has to be plain, as the data will be converted to data blob for the PIC32(!)
11:11
rajdeep
yes I'm trying to solve T1162
11:12
BAndiT1983
then please look at my other questions, as it's otherwise hard to guess what was done and why the image turns black
11:12
BAndiT1983
do you have the screenshot?
11:15
BAndiT1983
quick reboot, linux update requires it
11:15
BAndiT1983
changed nick to: BAndiT1983|away
11:15
rajdeep
the width and heights are right
11:17
rajdeep
I dont know what do you mean by format. But I had converted the png to extension ".h" with imagemagick
11:17
BAndiT1983|away
changed nick to: BAndiT1983
11:17
BAndiT1983
format like RGB565
11:17
rajdeep
What options do i need to check?
11:17
metal_dent[m]
are there any chances of GSoC getting cancelled due to corona?
11:17
BAndiT1983
of course! corona is spreading through email and phone! just joking!
11:18
rajdeep
lol
11:18
BAndiT1983
rajdeep: it's the task of the student to show how to solve the task, it's the reason we make challenge tasks to select the best students
11:19
metal_dent[m]
> of course! corona is spreading through email and phone! just joking!
11:19
metal_dent[m]
ofc not ð, just wanted to confirm!
11:20
BAndiT1983
any gathering could be cancelled, but as they are rather late in the year, like october, till then the emergency state should be over
11:20
BAndiT1983
but usually students don't travel for gsoc, so the possibility is rather low
11:21
metal_dent[m]
great then!
11:22
metal_dent[m]
btw do apertus hold any gathering/conference for GSoC students?
11:24
BAndiT1983
not really, as we are a rather small group and spread across the globe
11:24
BAndiT1983
but if you are near vienna you could use apertus hub
11:25
BAndiT1983
*-use *+visit
11:26
metal_dent[m]
this summer i might go to France so can visit!
11:28
BAndiT1983
hope till then the situation is over, been to paris recently and situation was ok, although less people, but it could happen that it will be declared to risk area and get bigger shutdown
11:39
metal_dent[m]
ohh! let's hope that doesn't happen
11:42
BAndiT1983
hoping same too, we've planned some trips this year, but at the moment it's not clear if german borders won't get shut for in and out traffic
11:44
metal_dent[m]
ohoo, in India there's no shutdown but my college has been suspended for a while
11:46
BAndiT1983
same happens here, but not in the whole country and city-wise clubs, bars and pubs close for a while
11:47
rajdeep
What color format is supported? When I select RGB565, i'm only getting bin files instead of c source header files
11:48
BAndiT1983
where do you select it?
11:49
metal_dent[m]
in the project description in "Goals" section, the last point is "..." so does that mean you'll add later or we have to suggest?
11:50
BAndiT1983
it was probably for the case i would come up with more goals, but had forgotten about it or wanted the first mentor (which is strangely absent) to add something
11:50
BAndiT1983
if you have good ideas, then you can extend your code or comment on the task in the lab
11:51
metal_dent[m]
okay, the other mentor is Supragya, right?
11:51
BAndiT1983
yes
11:51
BAndiT1983
i see his user here, supragya, but he hasn't responded the emails yet
11:52
BAndiT1983
also couldn't find much activities on his github account, but a week ago he created new repos
11:52
BAndiT1983
hope everything is ok
11:54
rajdeep
https://littlevgl.com/image-to-c-array
11:54
BAndiT1983
let me check, just create a small 100x100 png with alpha background
11:54
BAndiT1983
*created
11:55
metal_dent[m]
> i see his user here, supragya, but he hasn't responded the emails yet
11:55
metal_dent[m]
no worries, you and Bertl are quite helpful :)
11:57
BAndiT1983
the plan was that i guide the bootloader topic, as it's more advanced and Supragya as former apertus gsoc student the firmware
11:58
BAndiT1983
rajdeep: you could also try imagemagick, as i'm currently searching the online tool i'Ve used, gimp provides proper results also, but i don't like that it doesn't use HEX, but equivalent ascii or so, e.g. \377 or \011%
11:59
BAndiT1983
ah, here it is, if i'm not mistaken -> https://www.digole.com/tools/PicturetoC_Hex_converter.php
12:00
BAndiT1983
imagemagick would be preferable for repeatable results
12:00
BAndiT1983
but for tests the online tools are fine, as they usually also show the results instantly
12:02
BAndiT1983
just don't forget about 1-bit depth, as it's the goal of the task and DrawImage() uses full colour, so the data is quite big and clutters up the flash memory
12:03
rajdeep
digole doesnot work in India and brazil
12:04
rajdeep
Im switching to imagemagick again
12:06
BAndiT1983
alright
12:09
rajdeep
I think something is wrong
12:10
rajdeep
my dissected image shows that its dimension is 212x61
12:11
rajdeep
but after the converting, I had put the text into online character counter, there it shows that there are only 1657 ","(commas)
12:11
BAndiT1983
rgb565?
12:12
Bertl
rajdeep: sounds about right
12:12
rajdeep
i was expecting 12932 commas
12:12
BAndiT1983
don't forget / 2
12:13
Bertl
rajdeep: commas do not store bits :)
12:15
BAndiT1983
212 * 61 * 2 is the raw value for rgb565, as the reference
12:16
Bertl
raw what?
12:18
rajdeep
So I should expect 25864 commas?
12:19
BAndiT1983
212 px width * 61 px height * 2 bytes for colour -> this is raw for me
12:22
rajdeep
http://www.multipole.org/discourse-server/viewtopic.php?t=14391
12:23
rajdeep
it says RGB565 is 16bit
12:23
rajdeep
and depth is 16
12:23
BAndiT1983
but for the task it should be like 212 / 8 * 61, and / 8 is for 1 byte containing the data for 8 pixels
12:24
BAndiT1983
yes, rgb565 is depth 16, like 2 bytes, like 5 + 6 + 5 = 16 / 8 = 2 bytes
12:25
BAndiT1983
but this is what we already have, we need depth 1, like a stencil, so 1 would set a given colour for the pixel and 0 do nothing, maybe later we would use it for background, but not planned yet, as the background is provided by dialogue or the widget where the icon should be used
12:25
Bertl
off for now ... bbl
12:25
Bertl
changed nick to: Bertl_oO
12:32
BAndiT1983
metal_dent[m]: response to the last question on github is yes, let me copy my example for you, have started to add similar test for rectangle drawing, which creates it from 4 lines
12:32
BAndiT1983
https://pastebin.com/raw/aVt4WAfe
12:33
BAndiT1983
using // just for formwatting, otherwise it would concatenate the lines automatically
12:33
BAndiT1983
my framebuffer is 7x7 and rectangle would draw at 1,1 and size of 5x5
12:35
metal_dent[m]
and 0x7 in the array is?
12:35
BAndiT1983
just a number for colour, this one can be totally different
12:35
BAndiT1983
painter.DrawRectangle(1, 1, 5, 5, 0x7);
12:36
BAndiT1983
see last parameter
12:36
metal_dent[m]
okay!
12:36
BAndiT1983
ah, one very important thing i wanted to show -> uint16_t* framebuffer = new uint16_t[width * height]; memset(framebuffer, 0, width * height);
12:37
BAndiT1983
memset is necessary as array tends to have some junk data in it when it's freshly created, which is normal case when the application is not built for debugging, so all the values ahve to be zeroed
12:37
BAndiT1983
*have
12:38
metal_dent[m]
okay, will do this!
12:40
BAndiT1983
not the optimal way for now, but you can try to move it to some setup method, as it is something i've planned for the next iteration, still trying to fix text drawing and straightening out stuff
12:40
BAndiT1983
i mean the framebuffer and memset, something like SetupFramebuffer(...)
13:17
rajdeep
left the channel
13:18
rajdeep
joined the channel
13:19
rajdeep
Is there any way to check if image header is correct or not?
13:20
rajdeep
I'm not sure where the problem lies, the image header or because i'm using DrawImage()
13:21
BAndiT1983
DrawImage just grabs the bytes and copies them to framebuffer
13:34
BAndiT1983
you can also take 2 bytes and check if the value corresponds to right colour
13:38
metal_dent[m]
BAndiT1983: I updated my test, can you check and tell whether this is what you were expecting?
13:42
BAndiT1983
it's still a bit off, if you are drawing the icon with certain colour, then the expected output is the content of framebuffer you want to test against
13:42
BAndiT1983
do not use your logo data, but artificial input data
13:44
BAndiT1983
at the end you want to verify that it draws correctly, so you need to know the data which goes in and which you expect to land in the framebuffer
13:47
metal_dent[m]
> do not use your logo data, but artificial input data
13:47
metal_dent[m]
But as the method adds the color if bit is 1 so does it matter if I use logo data or artificial input?
13:48
BAndiT1983
the purpose of the test is to verify correct output at given input, so you need the artificial data, just imagine that the logo data will change in the future, then the test has to be adjusted over and over, not very convenient
13:49
BAndiT1983
look at the example of DrawRectangle() i'Ve posted before, create small framebuffer, like 16x16, create input data with 2x2 and expected output array, to compare against framebuffer
13:50
BAndiT1983
have done same thing for the rectangle yesterday, although i've fixed the problem beforehand, but the test could've told me about missing -1 for width and height, which i've just checked with GIMP and screenshot before the test
13:50
BAndiT1983
usually the approach is test-first, but we don't want to make the life more difficult for now
15:12
BAndiT1983
changed nick to: BAndiT1983|away
15:30
metal_dent[m]
okay!
15:30
rajdeep
left the channel
15:58
aombk
left the channel
16:01
aombk
joined the channel
16:02
BAndiT1983|away
changed nick to: BAndiT1983
16:04
metal_dent[m]
BAndiT1983: I think I did it!!
16:04
metal_dent[m]
the problem wasn't the code, it was the input image I was taking ð, I changed the input image to the apertus ring and the test passed!
16:06
BAndiT1983
please use artificial data, as the data used in the firmware can change
16:08
BAndiT1983
tests should always be independent of the application data
17:02
se6ast1an
does anyone have experience with manufacturing foam cushions with textile covers or custom surface treatments for foam pads?
17:03
BAndiT1983
se6ast1an: building cushion fort to protect from corona virus?
17:04
se6ast1an
yes, chinese manufacturer preferred :)
17:04
BAndiT1983
sorry, back to serious discussion, what is the target?
17:05
se6ast1an
https://www.softal.de/en/solutions/surface-treatment-utilizing-corona-technology/
17:05
se6ast1an
Surface treatment utilizing corona technology
17:05
se6ast1an
I want to design a shoulder pad for a rig for the Beta, like https://www.flickr.com/photos/camvatecom/24799989548
17:07
BAndiT1983
shouldn't be much different from the usual cushioning
17:07
BAndiT1983
when you have a foam in the basic form, then overlay it with the mesh fabric and stitch along edges
17:08
BAndiT1983
done only a bit of such things for the motorcycle seat some years ago, when it had to be slimmed down
17:09
BAndiT1983
from the picture it looks like they have 3 foam cubes with mesh over it, then folded into the bracket
17:12
se6ast1an
right, thanks, will check
17:14
BAndiT1983
you can also use the iron-on band which can hold the fabric together, used this to shorten long curtains, not a long-time solution, but for prototypes very good
17:15
BAndiT1983
maybe you know it, it's plastic which melts holds like glue when ironed over
17:17
BAndiT1983
*melts and holds
17:27
se6ast1an
I am more looking to understand an industrial process rather than a DIY approach
17:36
BAndiT1983
the industrial approach is like described before, they place 3 cubes of foam, span fabric over it and machine stitches along the edges of the foam, through the tension of fabric it smooths down the pads
17:38
BAndiT1983
also some forming is involved -> https://www.youtube.com/watch?v=mDh-1qEFqQk
17:39
flesk_
left the channel
17:39
alexML
left the channel
17:39
madonius
left the channel
17:39
alexML
joined the channel
17:40
madonius
joined the channel
17:40
flesk_
joined the channel
17:40
BAndiT1983
se6ast1an: probably something similar to this , but smaller -> https://www.youtube.com/watch?v=zsxM78KRg6U
18:03
metal_dent[m]
BAndiT1983: it's done, can you please check?
18:07
se6ast1an
BAndiT1983: does this surface finish look like hot treatment that just melts the foam into an even surface: https://cdn3.volusion.com/fmkpt.ufbwj/v/vspfiles/photos/fotga-shoulder-pad-2.jpg?v-cache=1411609171
18:20
se6ast1an
or is it molded in one go?
18:27
se6ast1an
seems it is https://image.made-in-china.com/2f0j00evbTaJPBEAgV/PU-Foam-Mold-for-Toy-Rugby-Football.jpg
18:28
BAndiT1983
se6ast1an: usually it's molded and "glued" together through heat, you can see the seam at the bottom
18:28
se6ast1an
right, thanks
19:13
BAndiT1983
https://www.youtube.com/watch?v=mXBGFkaQLes
19:30
felix_
left the channel
19:30
felix_
joined the channel
19:52
BAndiT1983
metal_dent[m]: the tests look better now, but i'm notsure we should name the files .xbm
19:52
BAndiT1983
will test your code tomorrow, as today i have to finish something else and of course work awaits tomorrow (at least till maybe home office is ordered because of corona outbreak)
19:55
BAndiT1983
for more straight formatting oyu could change the color value to something like 0x9, then output array would also have 0x0, then it would be easier to read
19:55
BAndiT1983
very good work so far
19:56
BAndiT1983
ah, one thing, no need to add d as variable, you can just calculate the index before the REQUIRE(), also REQUIRE could be placed at the end of method and only a boolean variable used, if != then break and REQUIRE would kick in and check if var is true or false
19:57
metal_dent[m]
> for more straight formatting oyu could change the color value to something like 0x9, then output array would also have 0x0, then it would be easier to read
19:57
metal_dent[m]
I just used the value off the top of my head ð
19:57
BAndiT1983
but this are small things and i will check on them tomorrow
19:57
ashok_singh[m]
Hello BAndiT1983, please review my task too.
19:57
metal_dent[m]
okay, thank you :)
19:57
BAndiT1983
no problem here with that, you could also use 0x00 for the other value
19:58
BAndiT1983
ashok_singh[m]: will try to do it tomorrow also, as i am doing something different at the moment
19:58
BAndiT1983
but i don't see any new commit since 15 days
19:58
ashok_singh[m]
<BAndiT1983 "ashok_singh: will try to do it t"> Okay it will be fine.
19:59
BAndiT1983
your loop also contains REQUIRE in it, not a good practice, one REQUIRE should it be at the end of the method and some var which contains the related value, depends on the test
20:00
BAndiT1983
wrong direction test is still not necessary, it was done for older code and will possibly removed later or kept to ensure no performance impact
20:00
ashok_singh[m]
> but i don't see any new commit since 15 days
20:00
ashok_singh[m]
Actually I squash commits for better review.
20:00
BAndiT1983
for newer code i expect proper placing of pixels
20:01
BAndiT1983
this is usually preferable when merging branches, but for feature development it's not that good, as it omits what was done and where possible problems came in, speaking out of experience in a bigger dev team
20:12
ashok_singh[m]
Yeah right, from next time I will keep that in mind and will do required changes soon in code.
20:19
ashok_singh[m]
Is there any chance that you guys thinking to remove push knob functionality in newer version of remote as it's not there in visualizer?
20:35
BAndiT1983
?
20:36
BAndiT1983
just because the knob is not attached again in the visualizer, it was working in former version, does not mean that we will remove it, nobody just had time to continue the development at full speed yet as the project is done mostly in spare time and everybody has their main job besides apertus
20:53
ashok_singh[m]
Oh right, I was just trying to list all possible events and button interaction.
20:54
BAndiT1983
12 buttons and clickable(!) knob for now, second knob will be removed as the purpose was not found, other buttons are not interesting for the control currently
21:34
se6ast1an
off to bed, good night
21:35
BAndiT1983
changed nick to: BAndiT1983|away
22:44
omar31
joined the channel
22:50
omar31
Hello Bertl, I have an inquiry regarding the eMMC task.
22:53
omar31
As I understand, the task is to implement the eMMC controller interface on the Zynq side
22:53
omar31
to send raw video data from the DDR memory via the 6 LVDS pairs
22:54
omar31
which will be used as medium speed channels
22:55
omar31
So, is this right?