#avr Logs

Sep 06 2019

#avr Calendar

11:15 AM _abc_: Is there a more electronics oriented channel on freenode besides ##electronics? I am fed up with the drugs and crazies circus there.
11:20 AM djph: not that I'm aware
11:20 AM _abc_: Ok, since this is ultimately an atmega8 project, I'll try here. Is anyone aware of a way to model a heated object (resistor + heat load) using RC circuit for electronic purposes? I found modeling info for buildings, it gets very complex. I do not care about sun load and 12 other things. And their models are not suitable for what I do. I have a working spice circuit that simulates what I need but I made it
11:20 AM _abc_: mostly by trial and error and I do not claim to understand what I did. Multi stage RC circuit.
11:20 AM _abc_: So any info on papers, or pages online discussing this would be welcome?
11:24 AM nohit: _abc_: #avrs
11:27 AM _abc_: Is there any known work on soft serial with "simple" api? I need to wedge in a receiver into an existing project, I did the tx easily, the receiver is a little harder, did it in the past already several times. Any projects one would recommend?
11:28 AM Fuchikoma: "Emil hates Rue. He and agents will be promoting their own channel #avrs "
11:28 AM Fuchikoma: heh
11:28 AM djph: _abc_: why're you using software serial?
11:28 AM _abc_: yes
11:28 AM _abc_: djph: :)
11:29 AM _abc_: because the relevant hw serial pins are busy doing something else in the existing layout
11:29 AM djph: "yes" is not an answer to "why"
11:30 AM Fuchikoma: I find it hard to believe there isn't a code example of bitbang serial RX
11:30 AM djph: there probably is. Software serial is just pretty freaking awful
11:31 AM _abc_: It is not djph
11:31 AM _abc_: Fuchikoma: there is not?
11:31 AM _abc_: I think there is. And the A<censed> soft serial library exists too.
11:32 AM Fuchikoma: No, I'm saying there certaily is I just never looked for one
11:32 AM nohit: Fuchikoma: im not promoting. he asked for electronics channel on freenode and i gave him the best one
11:40 AM * _abc_ should start #notavr :)
11:42 AM Emil: Fuchikoma: Consider joining #avrs :)
11:42 AM _abc_: I am on it, considering the above anyway.
11:42 AM _abc_: ##electronics seems to attract a lot of nonfunctional individuals.
11:46 AM Fuchikoma: It's cute how you complain about offtopic discussion while you post random bullshit links almost constantly :)
11:52 AM djph: Fuchikoma: people are weird :)
12:01 PM _abc_: Fuchikoma: yes but I do that usually without touching any drugs crazy issues and describing my (nonexistant) hemorrhoids or worse.
12:02 PM _abc_: Fuchikoma: Admiteddly I go to ##electronics for some bullshitting and letting off steam since there is no one left on there who knows what he is on about. And brisk sales via viral irc linking is rife.
12:04 PM * _abc_ notes one certain nick associated with that is on here too. And in #avrs
02:30 PM _abc_: Is goto label; still considered harmful in the context of gcc-avr ?
02:31 PM _abc_: Also can one expect a function which is void func(void) and single instance to be inlined and it's returns be optimized a bit by the compiler or is the situation bad there?
02:31 PM _abc_: goto used to break optimization, I was taught to avoid it.
02:31 PM LeoNerd: C lacks a few of the nicer control structures in some other languages; goto can often be used quite neatly in C
02:32 PM _abc_: Can be yes, but should it be used?
02:32 PM LeoNerd: Can be handy to break out of multiple loops at once (because C doesn't have named loops a.la. Perl)
02:32 PM LeoNerd: Can be handy to do unwind/cleanup on errors (because C doesn't have resource cleanup)
02:33 PM _abc_: Oh I know all about using while() as once goto less exitable loop, and also about using flags to pass state out to exit the needed number of loops. The sugar from "other languages" I can do without.
02:33 PM _abc_: Also error stacks/queues and so on. I've been coding in C other than learning it since circa 1990
02:33 PM _abc_: 80% embedded.
02:34 PM LeoNerd: Eh; using lots of flags for exiting loops adds a performance hit though
02:34 PM LeoNerd: a goto is literally just a jmp, so it's nicely clean and efficient
02:35 PM _abc_: Yes but the things it jumps out of cannot be reordered among other things.
02:35 PM LeoNerd: while(things) { for(more stuff) { goto finish; } } finish:; ... <== neat enough
02:35 PM _abc_: By the optimizer
02:35 PM _abc_: I am telling how it was in the 1990s :) What about now?
02:35 PM LeoNerd: Seems fine now
02:35 PM LeoNerd: I do all my AVR stuff using -flto anyway
02:37 PM Fuchikoma: Who needs optimisers when you do everything in assembly :D
02:37 PM _abc_: Is -flto not the default with -Os ? I need to read more about options.
02:38 PM LeoNerd: -flto is its own thing, I don't think it gets enabled with any of the -O bundles
02:38 PM LeoNerd: It's quite a change
02:38 PM _abc_: Sight
02:38 PM _abc_: -t
02:39 PM LeoNerd: -flto is the thing that allows optimisation -between- .c files, where without that it can't
02:40 PM LeoNerd: So lots of things like HAL functions become inlineable and so on. It's highly useful to be able to constfold into params to HAL functions, especially simple things like hardware initialisation
02:50 PM _abc_: Interesting. I will try it.
02:50 PM _abc_: What do you do to judge the effect of your compiler flags? Compare output size? Other?
02:56 PM _abc_: LeoNerd: ?
03:18 PM LeoNerd: Well, one time I had to because the code literally didn't fit if I didn't -flto
03:18 PM LeoNerd: But also the disassembly shows the effects
03:22 PM _abc_: Interesting
11:58 PM day__ is now known as day