#apertus IRC Channel Logs

off to bed now ... have a good one everyone!

berlt where you based at ?

ohh

as it's 08:30 here, then good morning :P

Bertl is just asynchronous to the rest, so his good night is different from us

<BAndiT1983 "oh the joys of student life"> 🤣

it supports some subset probably, as i don't care much for emojis, never checked the settings

hi

And collaboration would in any case be great

Something equivalent of a current GoPro, or similar.

(Android of Ubuntu Touch)

interesting

self driving bicycle? :)

not much other information available

BAndiT1983: https://pastebin.com/84nwy346 got this far.

same as yesterday

i mean the format

* so I think, 255 -> transparent and 46 -> black

as 2 bits can only hold these

yeah, but we need to map these values to 2-bit

* 46. 164. 250 and 255 are the only vals which occur in the array

i am using jpg

we wanted exactly this, afaik

any image format? why should we? SVG or PNG are enough

what is the shape for?

issue with PIL is that it can't change bit-depth to 4

you're using quantize and it's already good enough, in my prototype it gives me from 0-3 with right conversion steps

<BAndiT1983 "you're using quantize and it's a"> using command line right ?

i think pixel_export thing is doing something wrong

did you see the code I linked

thisi show i get the pixels, but it's possibly also not the best way -> im = np.array(image)

as my goal is SVG mostly, to have crisp data

will this solution be okay as it's a challenge task and not a actual issue ?

PNG is also possible, cairosvg is just giving me rasterized data, if the file is PNG it just omits the SVG loading/rasterizing step

what we want to see are the approaches and clean code, to be able to tell if people know how to solve complex tasks

just to mention it again, as my impression at every gsoc is, that students are only looking at the goal we expect: the goal is just a goal, but the code, approaches and collaboration are the real things we use for the evaluation, i can even include a student who hasn't reached the goal into considerations, if i see that he/she just lacks some knowledge but is eager to learn and be active

<BAndiT1983 "and the code tells me a lot abou"> :))))

focus on the approach, think about how we can use your solution in the future, e.g. in CI builds etc., this is some sort of what you will experience in a company when developing software

<BAndiT1983 "and yes, the code tells me a lot"> as I said, it's not complete, so don't judge yet xD

image.show() in PIL helps to check if the image is right, by the way

yeah, I know.

I've also been working on the image conversion, here is my progress this far https://github.com/MarkusEngsner/png2Cheader

Since that's what I have been using as sample pictures this far

How are most assets created for the Firmware? Are the sources mostly .svg?

usually we've used SVG, but it's negotiable, depends on which format provides better results, my preference is SVG and resizing/resampling to get smooth edges

result = src.substitute(d)

sure

I prefer a bitmap (tif, png, etc.) converted from vector like svg to already iron out some downsampling issues before conversion

but exporting as indexed format from gimp for the converter if fine IMO

BAndiT1983: the keywords, like WIDTH, HEIGHT etc

later one can supply additional values, which are not known beforehand, e.g. d['PIXEL_DATA'] = ','.join(str(im_data).format('X'))

I tried to use the template that was already present, without modifying it

maybe one can set identifiers for that too, but the original template came from other tests at first

as we've discovered that XBM is sufficient, we kept the idea of 2bit icons in the backlog, but the template could be useful with few adjustments or identifier settings

what the template lacks at the moment, is some field to tell if data is 2 bit or 1 bit, if we agree on using only 2 bits in the future, then it's ok, but if we want to go higher at some point, then it would be necessary

<BAndiT1983 "what the template lacks at the m"> more like include a comment there right ?

Ohk, change Icon struct inshort

*sorry for that many "checks", but in german it's easier to find different substitutes for words

PIL literally doesn't support that feature

> my conversion changes the image to grayscale, quantize() and maybe resize to do sampling for smooth edges

*nope

> you're using quantize and it's already good enough, in my prototype it gives me from 0-3 with right conversion steps

PIL

no problem, just look through the logs, as there are already quite many hints (maybe too many ;) )

hi satacker[m], welcome!

* Hello , My name is Shreyas Atre [github](www.github.com/SAtacker) . I am looking to participate in gsoc 2021 (student obv) . I have looked up the tasks and i think i can do at least task 1 of kernel challenge

hi satacker[m] and welcome

it seems like you're try to edit messages, but IRC doesn't support this -> http://irc.apertus.org/index.php?day=18&month=03&year=2021#162

Yup . I checked my github link and doesn't seem to work but fine

case closed! ;)

by "data sent to the kernel" means like echo "Hello" > /proc/some_module

oh right

SVG:s should now work fine too, https://github.com/MarkusEngsner/png2Cheader

nice! already seen the changes ;)

off for now ... bbl

hi rassu-123, yes, also for that, it's the main channel of apertus°

I am second year undergraduate. I have knowledge of C++,C ,python ,arduino,I love the combination coding +electronics. As a beginner I don't know much about electronics and things discussed in project, but i would be very happy if anyone can give me a small beginner task related to my electronics or something like arduino,rasberry pi,or

have you looked at available task and have some preference there?

as the USB VID/PID is not a real task, it involves just changing a couple of numbers in the firmware, so it's still there by mistake

okh ,thank you

https://github.com/VedantParanjape/image2array finished it, let me know if there's any nit picks

thanks, will check a bit later

hi hp_dopeman[m]

vedant16[m]: just as a hint, you can do some things much easier, e.g. array to string and byte concatenation, just check what numpy is capable of

<BAndiT1983 "what is the big comment in the m"> it's a doc string

<BAndiT1983 "vedant16: just as a hint, you ca"> I tried with array2string function

and since this can also be used as a library, it is useful

ohk, I'll remove it

BAndiT1983: apparently array2string wasn't working

I was getting ....... in between

template.py holds the template now

why template.py and not a separate .h or .txt??

I don't need to read files, I can just import it

really? and what happens when you write such a tool in C++? should the people build it every time they need to adjust the template?

hm, ok

with .txt I'll have to do it in code

I have changed the approach little bit, also updated the code and doc : https://github.com/Ddevendra/AXIOM-Remote/blob/2bitIconChallenge%40Devendra/Docs/Documentations/2Bit%20Image%20Support.md

<BAndiT1983 "nothing matters at all, you'll t"> It was a genuine question :(

we're using stubs at work to simulate REST endpoints, people started to place JSON responses in the stub script, this were long lines without formatting, as it was killed by intellij and such, so had to look for a better approach, luckily the framework supports external files, this allowed to use also placeholders for different answers, which previously was also done manually

seen same problems in quite some companies, learned to avoid some obstacles and possible problems, software development requires thinking ahead, especially in terms of maintenance, which is often done by colleagues

interesting, haven't seen cv2 approach yet

also interesting concatenation, but it's quicker and easier to do with shifting

as projects Java, but stubs were done in groovy at first, after i had to rework quite some stuff it ended in JSON, as the framework has features which people just haven't considered

<BAndiT1983 "as projects Java, but stubs were"> ohhhh

software partiallly from 1988, with macro hell from 2005 in C++ :P

<BAndiT1983 "software partiallly from 1988, w"> what's macro hell?

so my principle is KISS, even for complex software, as every problem can be broken down in manageable pieces, but far too often developers tend to write spaghetti code which causes even more trouble, that's why we do pull request reviews and static analysis a lot, also code coverage is a big point along with the related unit tests

<BAndiT1983 "so my principle is KISS, even fo"> Totally agree, I was writing the script while attending data science lecture lol, won't be lazy again 😰😰

ii = ...

that shit won't go into a commit, I assure you 😂

😂😂 OMG

he tried to save storage /s

or she*

that means sarcasm

but my favourite case was (in another company) to find methods in a Java class like get_2415(...), asked a colleague what it is, he said that it's the chapter 2.4.15 in the project specification book

and it really breaks me, when classmate commit code without even compiling 😂

as it was in the area of my module development had to rework it to getInboundItem() etc., before leving the company seen in the email that he suggested to remove my refactoring, case closed

I have added CI builds

the colleague told me following: "how often have you seen the specification book changing?", told him that it happened 5 times in the last 2 weeks by then, the chapters moved quicker than expected, as multiple people were writing specs

hahahaha

<BAndiT1983 "ah, you're using the github work"> yeah

satacker[m] is that guy😂 he's my junior actually

that's karma catching up to him lol

which can be futile, as people at work sometimes avoid to press the keys for code cleanup or don't even enable the checkbox for automatic process in intellij

BAndiT1983: If you want I can also create an account on my quasselserver for you...