Current Server Time: 04:53 (Central Europe)

#apertus IRC Channel Logs

2020/04/28

Timezone: UTC


00:06
Spirit532
left the channel
00:06
Spirit532
joined the channel
00:09
vup
ok not that much new from me this week
00:09
vup
anuejn did a lot of work on our nmigen gateware
00:09
vup
there is now a (low performance) axi lite interconnect and simulations / simulation helpers for the axi interface
00:09
vup
small example of how the new abstractions are used:
00:09
vup
https://github.com/apertus-open-source-cinema/nmigen-gateware/blob/master/src/connector_test.py#L42-L49
00:09
vup
also some work started on a high performance axi writer
00:09
vup
I did some small fixes of the micro r3 schematic (added some missing pullups) and continued filling the missing parts (adding decoupling and connecting the last few bits)
00:09
vup
Also some work done on nctrl, mostly some more internal improvements and the start of an actual test suite:
00:09
vup
https://github.com/apertus-open-source-cinema/nctrl/blob/afd2e725391a7a2a1169b16f67974c8625fd3185/src/test/mod.rs#L59-L95
00:13
Bertl
nice
00:14
Bertl
btw, I got all the stuff for testing the SD interface together (old interfaces as well as new gadgets to test with)
00:15
Bertl
haven't tested yet though, but planning to do so in the next few days ... what firmware does use the new FSBL?
00:20
vup
nice
00:20
vup
the new firmware hasn't been using fsbl for some time, (it uses u-boot spl instead)
00:20
vup
so any of the newer firmware builds should work
00:20
vup
newest currently would be https://github.com/apertus-open-source-cinema/axiom-firmware/releases/download/nightly%2F7bb337cf/axiom-beta-7bb337cf.img.xz
00:21
Bertl
okay, perfect, tx
00:21
vup
ok, off to bed now, have a good one!
00:21
Bertl
nn
00:35
Bertl
left the channel
00:36
Bertl
joined the channel
02:44
aombk
joined the channel
03:13
Bertl
off to bed now ... have a good one everyone!
03:13
Bertl
changed nick to: Bertl_zZ
05:09
BAndiT1983|away
changed nick to: BAndiT1983
06:29
BAndiT1983
changed nick to: BAndiT1983|away
06:42
se6ast1an
good day
06:48
BAndiT1983|away
changed nick to: BAndiT1983
07:11
metal_dent[m]
Hello!
07:25
se6ast1an
how is it going metal_dent[m]?
07:38
metal_dent[m]
going great, woke up a little late today ':D , will start working now..!
07:40
BAndiT1983
ah yes, the student life :P
07:49
mumptai
joined the channel
08:10
se6ast1an
well lockdown definitely makes it harder to get up in the morning :)
08:11
se6ast1an
at least for me
08:11
se6ast1an
the way to the office is too short...
08:12
metal_dent[m]
you don't have work-from-home?
08:16
BAndiT1983
we have, but some are still more students than others :D my alarm rings at 7am at latest and i start almost instantly my laptop after that
08:24
metal_dent[m]
XD
08:29
se6ast1an
my alarm goes off at 8 am but it takes me at least half an hour to arrive in the living room :P
08:36
BAndiT1983
https://giphy.com/gifs/crawling-l3aV3gXxYgS9q/fullscreen
08:39
BAndiT1983
but on the serious side, it's good to have a fixed time plan, as i can stop my work around 3PM or 3:30PM and do other stuff, although at the moment many things are not possible, so i can focus on 3d printing and software development more
08:41
metal_dent[m]
yeah, having a fixed time schedule is the best!!
08:47
BAndiT1983
have you implemented ImageButton already?
09:07
BAndiT1983
metal_dent[m]: se6ast1an: i would suggest to keep the images for the documentation in the repo and not somewhere in the cloud
09:08
metal_dent[m]
i guess maybe that's why they aren't appearing in the README..?
09:09
se6ast1an
yeah good idea
09:09
se6ast1an
there is https://github.com/apertus-open-source-cinema/AXIOM-Remote/tree/dev/Docs/Images/Structure
09:09
se6ast1an
for placing illustrations/images, etc.
09:11
metal_dent[m]
should I make an svg file too?
09:12
BAndiT1983
yes, please, have extracted it locally and started to adjust it a bit, as it was a bit off, but it's not finished yet
09:13
metal_dent[m]
okay then I'll wait for you to finish
09:16
BAndiT1983
nah, just push the version from the gdoc first
09:17
BAndiT1983
i will upload mine, when i have straightened a bit out, as it has produced too many objects
09:33
metal_dent[m]
okay, have added the png file
09:36
BAndiT1983
ehm, i was hoping for the SVG, as changeing the PNG is not that easy
09:36
BAndiT1983
*changing
09:37
metal_dent[m]
ohh
09:38
BAndiT1983
for docs and diagrams i would always prefer SVG first, if it's not a full color image example
09:41
metal_dent[m]
do you use Inkscape for conversion?
09:42
BAndiT1983
conversion of what?
09:42
metal_dent[m]
to svg image?
09:43
BAndiT1983
ou can download the SVG version directly from gdocs, just open the image and go to the menu in the diagram editor, there is download sub-menu and multiple image formats available
09:55
futarisIRCcloud
left the channel
10:36
metal_dent[m]
BAndiT1983: I created ImageButton class which inherits IButton
10:37
metal_dent[m]
also you recently suggested to not use switch-case for buttons then what can we use..?
10:46
BAndiT1983
there are a couple of examples in the code, which use button handlers, as i can't go fancy, they are plain static functions/methods
10:47
BAndiT1983
bigger question is, how to supply the data to them and in which format, should the screen be the sender or the button or both, as we for sure need to switch dialogs using them, so menu system should be supplied in some way and so on, many questions are still open
10:52
se6ast1an
what BAndiT1983 means I think is that its not something that can be solved in the same go as creating the ImageButton
10:54
BAndiT1983
exactly, this is totally separate problem to solve
10:54
metal_dent[m]
Okay and can you briefly describe the tasks of the ImageButton class? (want to make sure so that I don't go in the wrong direction like last time)
10:55
BAndiT1983
just as example and for further reading -> https://www.gamedev.net/tutorials/programming/general-and-gameplay-programming/effective-event-handling-in-c-r2459/
10:55
BAndiT1983
what is possible at the moment in the ImageButton class?
10:57
metal_dent[m]
afaik the drawing of the sidebar button images, implementation was the switch-case which i removed
10:58
BAndiT1983
ImageButton should only have one purpose, showing a button with given icon, i hope there is a possibility to provide the icon through constructor or setter method, although constructor should be the primary source
10:58
BAndiT1983
*-source *+choice
10:59
metal_dent[m]
only that? so the button handling will be in another class?
10:59
Bertl_zZ
changed nick to: Bertl
10:59
Bertl
morning folks!
11:00
BAndiT1983
Bertl: hi
11:00
BAndiT1983
metal_dent[m]: which button handling would you place in the button class?
11:02
metal_dent[m]
the button handling as in if we press button_7 we goto "Home" and like that...
11:02
BAndiT1983
?
11:02
BAndiT1983
and how do you want to have different handling if you place it only in one class??
11:03
metal_dent[m]
no no, i meant the "code" for the "handling"
11:04
BAndiT1983
it is already prepared and was working, you should check the code
11:04
metal_dent[m]
wait... i guess I'm getting confused ':D
11:11
BAndiT1983
metal_dent[m]: look at IButton, Activate() is not the best name yet, but this is the hinge of the whole handling
11:17
metal_dent[m]
yeah..
11:17
BAndiT1983
at the moment we need better glue than switch/case, namely adding handlers which have button ID and then the menu system would handle highlighting and other things
11:30
Bertl
metal_dent[m]: here you go: http://vserver.13thfloor.at/Stuff/AXIOM/REMOTE/remote_v0.14.pdf
11:33
metal_dent[m]
Bertl: thank you!
11:53
Bertl
np
12:13
max_bxl
joined the channel
12:13
max_bxl
hello everyone !
12:15
se6ast1an
hi maxime!
12:18
berto_bxl
joined the channel
12:18
se6ast1an
hey there
12:18
se6ast1an
how are things going?
13:05
BAndiT1983
changed nick to: BAndiT1983|away
13:12
max_bxl
se6ast1an we are testing the fil_reg 15 0 method to stop hdmi output
13:12
max_bxl
but we need to be root to do so
13:13
max_bxl
is there a way to write a script that can do it then ?
13:13
max_bxl
shall I copy cmv_reg for example ?
13:28
max_bxl
I created fil_reg.sh from cmv_reg.sh
13:28
max_bxl
and it works just fine
13:28
max_bxl
!
13:36
se6ast1an
yes root/sudo is required for any change to the camera parameters curently
13:36
se6ast1an
great, so what is the script you used now exactly?
13:48
max_bxl
https://pastebin.com/5uvaBD9W
13:50
se6ast1an
great, thanks
13:51
max_bxl
I wonder what is best practice for the axiom between #!/bin/bash or #!/bin/sh
13:51
max_bxl
?
13:52
se6ast1an
shouldnt matter
13:59
vup
max_bxl: I would say use /bin/sh if you don't need any bash features and /bin/bash if you do
14:01
max_bxl
ok thanks
14:01
BAndiT1983|away
changed nick to: BAndiT1983
14:02
vup
max_bxl: do the scripts work for you? then I will add them to the normal firmware builds
14:03
max_bxl
yep it just did
14:03
vup
ok
14:08
max_bxl
about raw2dng
14:09
max_bxl
I'm still not sure how to indicate which darkframe and/or dcnuframe file raw2dng should look at while converting raw12 to DNG
14:10
max_bxl
for now, I put the *.pgm files in the same folder as the raw12 I want to convert and run raw2dng from there
14:11
max_bxl
but running it twice on the same image, with the --no-dcnuframe --no-darkframe options output the same result
14:11
max_bxl
(visible when you contrast the image at maximum)
14:12
max_bxl
I'm wondering what I do wrong, since I didn't find explanation so far in the wiki
14:17
se6ast1an
is your dark frame called darkframe-xN.pgm
14:18
se6ast1an
where N stands for the gain
14:18
max_bxl
yep
14:18
se6ast1an
and does raw2dng output anything about what it is doing?
14:18
se6ast1an
can you upload raw12 and darkframes somewhere?
14:20
max_bxl
https://pastebin.com/MHF7XeUw
14:21
max_bxl
https://cloud.apertus.org/index.php/s/Lpo9f28GmsNeeTs in timelapse/20200323_rushes
14:21
se6ast1an
can you type "which raw2dng"
14:21
se6ast1an
thanks
14:22
max_bxl
gives /usr/local/bin/raw2dng
14:23
se6ast1an
raw2dng might be confused about the current path it is run from
14:23
se6ast1an
and looks for darkframe there...
14:23
se6ast1an
in which directoy to you have darkframe currently?
14:24
max_bxl
ok, it was copied there during "make install"
14:24
se6ast1an
you do this on the beta or your pc?
14:24
max_bxl
same directory as in the cloud
14:24
max_bxl
but on my computer
14:24
max_bxl
on the pc
14:25
se6ast1an
ok, please try copying the raw2dng executable to your current work directly
14:25
se6ast1an
cp /usr/local/bin/raw2dng /home/maxime or whatever your workdir is
14:25
se6ast1an
and then run "./raw2dng image00001.raw12" instead of "raw2dng image00001.raw12"
14:27
max_bxl
output (text and DNG) is the same
14:29
max_bxl
it's uploaded on the cloud "image00001.DNG"
14:30
max_bxl
any idea ?
14:30
se6ast1an
alexML: you there?
14:32
se6ast1an
downloading files and compiling raw2dng locally currently to verify
14:35
max_bxl
ok thanks !
14:36
se6ast1an
ok I can confirm that behaviour
14:36
se6ast1an
does the same here
14:36
se6ast1an
will check with alexML
14:36
max_bxl
what are the options?
14:36
max_bxl
ok thanks!
14:36
max_bxl
gotta go for an hour, will check the logs!
14:41
se6ast1an
right
14:49
se6ast1an
max_bxl: I think I have an idea what could cause the problem
14:50
se6ast1an
you did not use the -r dump sensor registers parameter when you captured the images
14:50
se6ast1an
so there is no metadata and raw2dng does not know which gain was used
14:51
se6ast1an
off for a bit
15:19
Bertl
off for now ... bbl
15:19
Bertl
changed nick to: Bertl_oO
15:29
berto_bxl
left the channel
15:33
se6ast1an
Should be easy to fix btw
15:48
berto_bxl
joined the channel
15:57
max_bxl
se6ast1an thanks!
15:57
max_bxl
will try asap
15:57
max_bxl
bye everyone
15:57
max_bxl
left the channel
16:12
berto_bxl
Hello Sebastian, everyone
16:12
berto_bxl
I'm starting RCN calibration from begining
16:14
berto_bxl
I'm having a little issue when taking darkframe, It stops taking full resolution image starting image 48
16:14
berto_bxl
https://pastebin.com/AaktXdSy
16:15
berto_bxl
I'm following instruction from firmware(2.0) https://wiki.apertus.org/index.php/Factory_Calibration_(firmware_2.0)
16:25
se6ast1an
Are you running out of space?
16:26
berto_bxl
just found out the solution
16:26
berto_bxl
yes indeed
16:52
panintended
joined the channel
17:03
metal_dent[m]
umm... shouldn't this be "highlighted" -> https://github.com/MetalDent/AXIOM-Remote/blob/dev/Firmware/UI/Widgets/PushButton.h#L13 ?
17:04
BAndiT1983
yes, looks like a typo
17:05
metal_dent[m]
it's used in MainPageButton and PushButton files
17:14
BAndiT1983
what is so special about a typo? it was a copy/paste thing between the 2 classes
17:15
metal_dent[m]
nothing special.. I was just looking at the button classes and found it
17:15
metal_dent[m]
do we need to fix the typo?
17:15
BAndiT1983
no, we will keep it to please the gods...
17:16
BAndiT1983
of course it should be fixed
17:16
metal_dent[m]
okay XD
17:21
BAndiT1983
by the way, highlighting should also move to the IButton class for now, usually i would say that we should create Button class, which is derived from IButton and PushButton should derive from Button, but for now i don't want to clutter up the system too much
17:21
BAndiT1983
so i'm cheating a bit and using the interface class as base class for the widgets
17:22
metal_dent[m]
yeah, I'm also making the ImageButton like the PushButton
17:22
metal_dent[m]
the idea of the Button class from IButton is good but will need to clean up a lot
17:23
BAndiT1983
what i've described is the usual way: interface -> base class -> derived classes
17:24
BAndiT1983
but we will do it afterwards, first we need progress and i can assist by refactoring, when you have more experience with OOP, then you can also start to straighten things out
17:25
metal_dent[m]
i just grep and checked out, there aren't much "hightlighted"s
17:25
BAndiT1983
as we are on a microcontroller level, i won't force full-blown OOP solution for firmware, so shortcuts are ok, but clean code is still mandatory, like short methods and clean structure with single reposnsibility
17:26
BAndiT1983
i know, as this 2 classes were written by me
17:27
BAndiT1983
will start cooking, but am still able to chat, have to think about the debug painter, have a solution, which is maybe not that beautiful, but would spare us quite some trouble
17:29
metal_dent[m]
great! and what's cooking? ;)
17:33
BAndiT1983
baking some vegetables in the oven (grill mode), like aubergines, zucchini and carrots, i slice them before, apply oil and spices, afterwards i place them in a bowl and pour vinegar over it and squeeze garlic over it, then it can be stirred and let it soak a bit before eating
17:36
metal_dent[m]
sounds delicious (and healthy)
17:39
BAndiT1983
not always, but often, one of the reasons is, that because of the fire we had no electricity for 1.5 days, so gave all the things from the freezer to my brother before it unfroze and gone bad, just have to stock it up again
17:46
panintended
Hi everyone o/
17:50
panintended
BAndiT1983: if you want some feedback (or even to setup a meeting to brainstorm) for the debug painter do let me know. If I can schedule it in advance, I should be able to put 2-3 hours aside.
17:52
panintended
*be it the debug painter or anything else
17:52
BAndiT1983
hi panintended
17:54
BAndiT1983
thanks for the offer, will possibly come back to it, just have also other things besides apertus which consume time, so my development efforts are a bit sporadic
17:54
panintended
I feel that :)
17:54
panintended
I have a similar sistuation myself
17:55
panintended
just know that the offer is there. It can even be a public Jitsi meeting that anyone can join
17:56
BAndiT1983
usually we do spontaneous meetings here, not many people are active so it's good that some life is happening on this channel besides the cosmic silence
17:56
BAndiT1983
also reading logs often when not online, just replying is difficult, but will install IRC on my smartphone
17:58
BAndiT1983
any news on the knob button? what is your general opinion on the structure of the code?
17:59
panintended
You can use a matrix bridge for that since freenode (users,channels, etc.) is bridged. Here: https://github.com/matrix-org/matrix-appservice-irc/wiki/Guide:-How-to-use-Matrix-to-participate-in-IRC-rooms
18:01
BAndiT1983
don't have matrix account, as i'm not using messaging that much
18:01
panintended
Actually the link is to setup of an IRC bouncer of sorts. But you can also just directly communicate from your matrix.org handle with freenode usesrs/rooms
18:01
panintended
ok, understood
18:02
BAndiT1983
my IRC account is configured on the apertus bouncer, have just to enter the credentials on my phone
18:02
panintended
ok, easy then :)
18:03
panintended
regarding the knob: I'm about to test what you'd sent (https://github.com/ocornut/imgui/issues/1889). I will put some time aside tomorrow as well, as today again my day was full
18:04
BAndiT1983
take your time, there is no urgency
18:05
Roopar
joined the channel
18:05
BAndiT1983
luckily my sprint change is next week, otherwise i would also have full day of phone conference, which drains the brain over 5 hours
18:05
Roopar
left the channel
18:06
panintended
I do have questions coming up as I'm going through the code (about the implementation itself and also about the direction things are going). I guess I should start writing things down :P
18:07
BAndiT1983
ask anytime, you can also send emails, this ones i can answer much quicker when not available here
18:08
panintended
thanks, appreciated
18:19
se6ast1an
just pushed a tiny improvement to raw2dng: https://github.com/apertus-open-source-cinema/misc-tools-utilities/commit/3ca6468e99c7915f391b8c45fae4e1a4738f61f7
18:19
se6ast1an
this should make it easier to spot the problem maxime ran into early
18:26
metal_dent[m]
BAndiT1983: fixed the typo!
18:31
BAndiT1983
metal_dent[m]: thanks, merged both PRs
18:38
metal_dent[m]
thanks!
20:40
vup
Bertl_oO: did you ever test the axi hp writer without multiple in flight transactions? If we didn't mess up anything, we are getting about 1436 MB/s without having multiple transactions in flight...
20:41
Bertl_oO
nah, never tested it out of the original design
20:44
Bertl_oO
but for writes, it shouldn't be too problematic to not have the addresses in advance
20:44
vup
ok, does ~1400MB/s seem reasonable for one writer?
20:44
Bertl_oO
I presume the memory controller will rearrange the transactions anyway
20:45
Bertl_oO
let me dig out the test values we got ... will take a few minutes
20:45
vup
sure
20:45
vup
they also seem to hint at some reordering / latency hiding in the axi hp interface itself in the zynq trm
20:52
Bertl_oO
so, a qick check gave we tested 11Gbit/s for a single reader/writer
20:52
Bertl_oO
which matches your results quite nicely
20:58
vup
ah nice
21:00
Bertl_oO
note that the DDR memory maxes out after the second writer IIRC
21:05
vup
yeah we can try that later
21:07
Bertl_oO
currently testing the sd-card boot
21:10
Bertl_oO
long cable is already enough to make the sd card fail in both cases, old and new firmware
21:11
vup
hmm too bad
21:11
vup
so maybe its the bootrom failing and not the fsbl
21:12
Bertl_oO
maybe, maybe it is just the controller
21:12
Bertl_oO
I'll investigate a little more there ...
21:13
Bertl_oO
should have tools now which were not available back then :)
21:13
vup
I thought it worked in u-boot / linux after booting from something else...
21:13
BAndiT1983
changed nick to: BAndiT1983|away
21:13
Bertl_oO
well, yes, the card 'works' once Linux is up
21:13
Bertl_oO
I can test that when booting from flash
21:14
vup
well then the controller shouldn't be the problem (if it is not misconfigured, but that seems unlikely)
21:14
Bertl_oO
I suspect a timing issue, but maybe it is a voltage level problem during detection
21:16
Bertl_oO
I'm currently using this for the 'long cable' test: https://www.adafruit.com/product/3688
21:18
vup
do you get anything on serial?
21:18
se6ast1an
off to bed, good night
21:18
vup
nn
21:19
Bertl_oO
nope, just the red led goes on, nothing on the serial
21:19
Bertl_oO
and I can't reset the device from that point on either
21:19
Bertl_oO
i.e. I have to power cycle for another test
21:20
vup
hmm that doesn't sound good
21:20
vup
maybe 68cm is a bit too much
21:20
Bertl_oO
which is the same behaviour I get when booting without SD card
21:20
vup
yeah
21:20
vup
did you ever try to determine the state the cpu is in via jtag?
21:23
Bertl_oO
nope, but I have jtag connected and plan to test :)
21:23
Bertl_oO
I'm also doing some tests with the SDIO mux I used to figure out whether it behaves differently from the card
21:24
Bertl_oO
it might be that we are dealing with two unrelated issues here
21:24
Bertl_oO
i.e. cable being too long and SD-mux behaving too unusual for the picky sd controller
21:24
vup
yeah could be
21:25
vup
did you measure the voltage drop accross the cable?
21:25
Bertl_oO
not yet, but this time I have the necessary instrumentation for this kind of checks
21:27
vup
nice
21:27
Bertl_oO
btw, I would suspect the JTAG to be disabled during the ROM bootloader no?
21:28
Bertl_oO
but we should know soon, just have to do some modifications to simplify power cycling first
21:36
vup
hmm yeah didn't think of that
21:36
vup
lets hope not
21:40
Bertl_oO
hmm, so with normal testing of the sd-mux, I only get 11.5MB/s thoughput on the card, while without I get 22.5MB/s
21:40
Bertl_oO
which means that at least something is off here
21:48
vup
i suppose the bandwidth is the same on a different "normal" pc?
21:48
Bertl_oO
yes, I'm always testing on both, a card reader and the MicroZed
21:49
Bertl_oO
it looks like the reference design does use much higher pullups and no pullup on the clock
21:49
Bertl_oO
I'll try that just for fun :)
21:49
vup
yeah that could be something
22:07
panintended
left the channel
22:20
Bertl_oO
well, the results are different now, not really better though
22:21
Bertl_oO
changed the pullups from 10k to 100k on CMD, DAT* and removed the pullup from CLK
22:21
Bertl_oO
power cycling the MicroZed now gives no red led
22:21
Bertl_oO
pushing the reset button activates the red led
22:21
Bertl_oO
no output on the console though (tested with both old and new firmware)
22:22
Bertl_oO
sd-mux still only achieves 11.5MB/s throughput
22:44
futarisIRCcloud
joined the channel
23:14
vup
hmm interesting