IRC Logs for #crux Monday, 2015-11-02

*** kori has quit IRC00:46
*** john_cephalopoda has quit IRC00:52
*** ivs has quit IRC00:56
*** kori has joined #crux00:58
*** madvlad has joined #crux01:03
*** wuyin has quit IRC01:28
*** nurplepurple has quit IRC01:28
*** nurplepurple has joined #crux01:30
*** wuyin has joined #crux01:30
Romsterskrzyp send me the changes to the maintainers of them ports.01:37
*** wuyin has quit IRC01:38
*** nurplepurple has quit IRC01:39
*** nurplepurple has joined #crux01:40
*** wuyin has joined #crux01:40
*** wuyin has quit IRC01:51
*** nurplepurple has quit IRC01:52
*** frinnst has quit IRC02:03
*** frinnst has joined #crux02:36
*** frinnst has quit IRC02:36
*** frinnst has joined #crux02:36
*** jusplainmike has quit IRC02:43
*** jusplainmike has joined #crux02:45
koriRomster: yo, are you getting this with webkit-gtk2?02:51
koriconfigure: error: Compiler GCC >= 4.7 or Clang >= 3.3 is required for C compilation02:51
tired8904.3 final is out..03:28
tired890tempted to see if IOMMU is fixed in this or I'll have to stay with 4.1 still..03:28
*** mavric61 has quit IRC03:32
*** mavric61 has joined #crux03:33
tired890good news lads!03:46
tired890In other news, I got excited about this: as a future upgrade for my system (say in 2 months time)03:48
tired890but unfortunately it needs a EFI bios for it to be bootable.. and my mobo is only legacy..03:49
*** deivior has quit IRC03:49
Worksterkori, on crux 3.1 nope it works here. built package for convince
*** xeirrr has joined #crux04:03
koriWorkster: well it didn't build with that package04:16
koriI'll have to wait for you to compile webkit or icecat with 3.2 packages04:16
koribecause neither are building here04:16
rmullTrying kernel 4.3 out now04:20
rmullInterestingly, now my noscript button doesn't hang my browser04:21
tired890rmull, I compiled and installed. But cant reboot yet as I have some work processes running04:21
tired890rmull, what noscript button?04:21
xeirrrrmull: you can add pf and ck patches to kernel then recompile.04:23
rmulltired890: there is a firefox plugin called 'noscript' to selectively disable javascript on webpages. When I used to click the button to show its menu, firefox would hang and be mostly unkillable.04:24
tired8904.2 has virtio gpu driver (modesetting for qemu/kvm guests).. but I never got to use this as AMD-Vi/IOMMU went haywire with anything in 4.2.x family..04:25
tired890rmull, ic04:25
tired8904.3 has lots of nouveau (nvidia open source driver) issues sorted also for the nvidia users among u04:26
*** SiFuh_ has joined #crux06:09
*** SiFuh has quit IRC06:11
*** jusplainmike has quit IRC06:18
*** jusplainmike has joined #crux06:18
*** xeirrr has quit IRC06:27
*** tired890 has quit IRC06:32
*** tired890 has joined #crux06:35
*** tired890 has quit IRC06:37
*** tired890 has joined #crux06:40
*** hhhhhhhh has joined #crux07:34
*** xeirrr has joined #crux07:51
*** hhhhhhhh has quit IRC07:58
*** hhhhhhhh has joined #crux08:41
*** hhhhhhhh has quit IRC08:51
*** SpacePenguin has quit IRC09:07
*** SpacePenguin has joined #crux09:08
*** Lukc has quit IRC09:11
*** Lukc has joined #crux09:12
*** pejman has quit IRC09:17
*** Lukc has quit IRC09:24
*** pejman has joined #crux09:29
*** Lukc has joined #crux09:37
*** cruxbot has joined #crux09:40
*** hhhhhhhh has joined #crux09:45
*** onodera has joined #crux10:29
*** BitPuffin|osx has quit IRC10:44
cruxbot[contrib.git/3.1]: pmwiki: update to 2.2.8110:51
*** xeirrr has quit IRC10:52
*** xeirrr has joined #crux10:55
*** pal_bth has joined #crux11:06
*** ivs has joined #crux11:08
*** heroux has joined #crux11:23
*** heroux has quit IRC11:31
*** tired_laptop has joined #crux11:48
*** heroux has joined #crux11:49
*** _ixxi_ has quit IRC11:49
*** tired_laptop has quit IRC11:51
*** penguinfan has joined #crux12:01
*** ivs has quit IRC12:43
*** BitPuffin has joined #crux12:56
*** xeirrr has quit IRC13:00
*** tired_laptop has joined #crux13:22
*** tired_laptop has quit IRC13:50
*** jusplainmike has quit IRC14:43
*** jusplainmike has joined #crux14:45
*** gb00s has joined #crux15:12
*** pal_bth_ has joined #crux15:17
*** _mavric61 has joined #crux15:28
*** tierd891 has joined #crux15:29
*** erdic_ has joined #crux15:30
*** nwe_ has joined #crux15:31
*** tired890 has quit IRC15:35
*** mavric61 has quit IRC15:35
*** nwe has quit IRC15:35
*** delisa has quit IRC15:35
prologicjaeger:  ahh cool! We're actually living in Palo Alto :)15:37
*** gb00s_x has joined #crux15:42
*** penguinfan_ has joined #crux16:22
*** penguinfan has quit IRC16:22
*** xor29ah has quit IRC16:23
*** penguinfan_ has quit IRC16:24
*** gb00s_x has quit IRC16:28
*** gb00s has quit IRC16:29
rmullMan, I suck at makefiles16:35
*** delisa has joined #crux16:55
*** gb00s has joined #crux17:15
cruxbot[core.git/3.1]: util-linux: update to 2.27.117:22
*** gb00s has quit IRC17:40
z3brarmull, need a hand?17:48
cruxbot[core.git/3.1]: grep: update to 2.2218:03
*** gb00s has joined #crux18:24
*** ivs has joined #crux18:35
*** pal_bth_ has quit IRC18:46
*** xor29ah has joined #crux18:46
*** BitPuffin has quit IRC18:49
skrzyp XDDD18:57
rmullz3bra: Sure. Let me try to compose a question that makes sens.19:00
rmullI have a flat dir of source files, *.c and *.h. I have an obj/ dir that I want to keep my object files in. I have a test/ dir that I have small .c files for testing various routines in their parent dir. Here is the Makefile I have at the moment:
rmullI am having trouble with the test target - I can't seem to get gcc to see the object files that contain the code to be tested19:03
rmullI tried using TESTOBJ to remove main.o, since my test files all have their own main() routines19:03
rmullCurrently, the test C files get compiled without being able to see the .o files that contain the code to be tested, and I gave up so I could do some actual work :\19:05
rmulldang tests.19:05
tilmanrmull: i'll trade you makefile help for electronics help19:14
rmullHow low level? I'm not an analog guy19:18
rmullBut I can take a glance19:18
tilmanlet me see if i can fix your makefile first19:19
rmullMuch obliged :)19:21
tilmanyour makefile is making me angry19:24
rmullMe too19:24
rmullDoes the organization of the code make sense?19:24
tilmando you want to end up with _one_ binary that contains all of your tests?19:24
rmullMany binaries, each with their own main()19:24
rmullbut all stored in test/19:24
tilmanbut you only have one 'test' target in the makefile?19:25
tilmanwhich ony builds one binary?19:25
rmullThe goal would be to build each of the tests in test/19:25
rmullIt may look like I'm only building one binary now, but that would be a mistake19:25
tilmando you want test/sadness_test.c to be linked as obj/sadness_test ?19:26
tilmanor where do you want the binaries?19:26
rmulltest object code can end up in test/obj/ - the test executables can be with their source files in test/19:26
tilmansome observations:19:27
rmullThough I'm not too picky about the test object code, it doesn't need to be stored in an obj/ dir19:27
rmullDisclaimer: I haven't ever bothered with test code before19:27
tilmanseparating object files from source, but not separating binaries from source is weird to me19:27
tilman"gcc -o $<" looks backwards19:27
tilman$< is your first dependency19:27
tilman$@ is the output19:27
tilmanso usually you'll want:19:27
tilmangcc -o $@ $<19:27
tilman(or $^ which is _all_ deps)19:27
tilmanrmull: let me give it a whack19:28
rmullI'd settle for advice,  you don't need to solve the whole thing for me (unless you want the practice)19:28
rmullThanks for the tips so far19:28
tilmani've never seen "test: CFLAGS += ...." btw -- does that work?19:29
tilmansetting variables like that/19:29
rmullIt might have worked, because I do see the CFLAGS in the gcc command when it gets run19:30
rmullBut the goal here was to append some additional #defines when the test target was invoked19:30
rmullso there may be a better way19:30
rmullThis is a frankenstein copy/paste and guess/check Makefile at the moment19:31
tilmanshould "make test" run the tests as well?19:31
tilmanand can i use a 10K potentiometer on a 3.3V input and get voltages > 1.0V as output?19:32
tilmanwith the pot at the positive end, and 3.3V input, should i measure 3.3V as the pot's output?19:32
rmullmake test doesn't need to run the tests yet, I think eventually I may make it call a shell script that will run them19:33
tilmani'm not talking about need, i'm talking about want/like/wish19:33
rmullAh - yes19:33
tilmani think you do want that19:33
rmullIs the pot part of a voltage divider, or is it just 3.3V rail -> pot -> input pin?19:34
tilmanthe latter19:34
tilmanwhere my "input pin" is just a multimeter19:34
tilmanbut i only get it to read voltages in the mV range19:35
tilmani have verified that i _do_ have 3.3V connected to the pot though19:35
rmullThat's normal - if you want to create a variable voltage that is adjustable with a pot, you'll want to use another resistor and make a divider19:35
tilmanokay, thanks19:35
rmullRight now you're just making a pullup resistor of variable resistance19:36
tilmane131 is your production binary, right?19:37
tilmanimo the test target shouldn't depend on it, but ymmv19:38
rmullYes, that's the production binary- I was hoping to make the test target dependent on its objects19:38
tilmanrmull: i only added/changed the final two targets19:38
rmull(except for main)19:38
tilman(not counting clean)19:38
tilmanwhat i usually do is have:19:38
tilmanAPP_OBJECTS := untestable_app_code.c19:39
tilmanTEST_OBJECTS := test_code_only.c19:39
tilmanCOMMON_OBJECTS := stuff_for_both_app_and_tests.c19:39
tilmanwell, i mixed up .c and .o there but you get the idea19:39
tilmancurse my lack of electronics equipment19:41
tilmani don't even have a bag of resistors19:42
rmullDo you have an MCU? Often they feature internal pullups on the GPIO inputs, so if you could turn one of those on, you could use the input pin with the pullup on and a pot to ground to make a crappy divider with crappy specs19:43
rmullOr you could salvage one pretty easily from any electronic gadget19:44
tilmani think i'll pass and use a configurable PSU at work tomorrow19:45
rmullI am having trouble with the test target after a clean - make: *** No rule to make target 'test/e171_test', needed by 'test'19:46
rmullIs this because the test code isn't included in the OBJ wildcard search?19:47
tilmanmmh, works for me19:48
tilmani'll tar up my directory19:48
rmullThanks, downloaded19:49
rmullIt doesn't seem to work as expected with multiple test source files19:53
rmullI added a prod2_test.c and got the error I'm getting now19:53
tilmantest/prod2_test.c would require prod2.c19:54
tilmanwasn't that the intention?19:54
rmullYes, hold please, let me confirm19:54
tilmanone test module for each production code module?19:54
rmullYes, you're correct19:55
rmullI had a typo in a filename in my tree :)19:56
rmullIt totally works19:56
rmulltilman: Thanks much, man.19:56
rmullI am excited to finally be writing tests for my code19:57
tilmanalso check out check (as in opt/check) and libcunit for small unit test frameworks19:57
tilmani think i like check a little better19:57
tilmanif you're open to writing your tests in c++, look at google test19:57
rmullI will, thanks. Let me know if there is anything else I can do to help out electronics-wise.19:57
tilmanalso, this guy
tilmanblogs/blogged about unit testing c code for MCUs19:58
tilman(see the 'top posts' on the right)19:58
tilmanlet me know when you need advice on the topic in general19:59
rmullI appreciate it19:59
tilmana year ago i unit tested an xc164 program (c) with google test on msys in windows19:59
tilmananother thing20:01
tilmanwhat the what?20:04
rmullI think you just provided me with like a solid week of reading20:05
*** gb00s has quit IRC20:28
rmulltilman: The test target in the Makefile seems to fail when a module-under-test is dependent on another source module because the call to gcc only includes the module-under-test, and not its dependents20:44
rmullAny tips to work around this?20:45
tilmandefine separate targets to build the test binaries and list the dependent obj files manually20:45
tilmanwhich sucks, but. :|20:46
rmullYeah, hmmm20:46
rmullThat's probably the easiest path forward for now20:46
tilmanpulling in _all_ object files will typically break for this kind of setup, since your test code will likely define functions that are also in $production code20:47
rmullThat's a good point20:47
rmullI'm sure there must be some templates for a development setup similar to this one that I have in mind20:47
rmullI probably just made some minor incorrect assumptions to start with20:48
rmullopt/check seems to be recommending autotools - is that what you prefer as well?20:49
*** gb00s has joined #crux20:50
rmullIn general it strikes me that all of the test "frameworks" include a good deal more features than I am interested in20:51
tilmanjust build with -lcheck and you're done20:53
tilmanno need for autotools (at least in your project)20:53
*** gb00s has quit IRC20:55
tilmanrmull: if you compile all of your production code into libfoo.a, and then link each single test program against that lib you might get away without listing the dep objs manually21:03
tilmanbecause then the linker would (should?) only pick the objects that it still needs symbols from21:03
*** onodera has quit IRC21:06
*** onodera has joined #crux21:06
z3brarmull, I'm back sorry21:23
z3brayou have one test per C file?21:23
z3bra("find . -type f" might help me figure out what's happening)21:24
*** deivior has joined #crux21:36
rmullz3bra: Thanks anyway, tilman sorted me out22:07
rmullWhy do some style guides (like openbsd's) suggest putting the return type of a function declaration on its own line?22:22
rmullNo matter what I do I can't think of a reason why that would benefit me22:23
rmulltilman: I decided to simplify everything. Instead of having separate test binaries, I have a test function for every function that is conditionally compiled and called in main, and each function is stored with the function it's testing22:36
rmullThe only difference between my Makefile targets is now just a #define or two22:36
*** onodera has quit IRC22:57
cruxbot[xorg.git/3.1]: libvdpau: corrected url23:04
*** hhhhhhhh has quit IRC23:25
skrzypSpeaking to PC World, Microsoft Corporate Vice President Joe Belfiore explained that Windows 10 is constantly tracking how it operates and how you are using it and sending that information back to Microsoft by default. More importantly he also confirmed that, despite offering some options to turn elements of tracking off, core data collection simply cannot be stopped:23:42
skrzyp“In the cases where we’ve not provided options, we feel that those things have to do with the health of the system,” he said. “In the case of knowing that our system that we’ve created is crashing, or is having serious performance problems, we view that as so helpful to the ecosystem and so not an issue of personal privacy, that today we collect that data so that we make that experience better for23:42
*** hhhhhhhh has joined #crux23:45
*** wuyin has joined #crux23:50
*** hhhhhhhh has quit IRC23:50
*** nurplepurple has joined #crux23:51

Generated by 2.11.0 by Marius Gedminas - find it at!