| --- Log opened Sun Jul 09 00:00:15 2017 |
| 00:00 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 00:00 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 00:02 | <&McMartin> | The part I have the most trouble with in that article is actually at the very beginning; I'm unclear as to why the initial empty template is necessary |
| 00:03 | <&McMartin> | Otherwise I follow it fine, which may not say much... |
| 00:03 | <&McMartin> | ... because the codebase I've been working on for the past 2 years includse a class very like your event broadcaster, written by me, early on in that 2 years. |
| 00:03 | <&McMartin> | So that code looked fine~ |
| 00:05 | < RchrdB> | ...has anyone tried programming on a device that resembles an MS Surface? |
| 00:05 | <&McMartin> | Surface or Surface Pro? |
| 00:06 | | * RchrdB looks up the difference |
| 00:06 | <&McMartin> | "The one that actually runs Windows", which includes Surface 3 on |
| 00:06 | <&McMartin> | As opposed to the one running Windows RT on ARM. |
| 00:06 | <~Vornicus> | I've done some on a surface. the on screen keybaord is big enough to actualy use as a keyboard but is missing all the good nav buttons |
| 00:06 | < RchrdB> | does Win10 run on ARM these days? |
| 00:07 | <&McMartin> | I have run MSVS2012 on a Surface Pro 3 with an attached keyboard before. |
| 00:07 | <&McMartin> | But mostly as a stunt |
| 00:07 | <&McMartin> | But it was able to chew through a very sizable project. |
| 00:08 | <&McMartin> | But the Surface Pro 3 is Actually A Laptop. |
| 00:08 | < RchrdB> | My current laptop is somewhat fucked; keyboard's terrible and it has atrocious cooling problems such that its fans sound like an aircraft taking off pretty much all the time it's switched on |
| 00:08 | <&McMartin> | And to my knowledge, Win10 itself does not run on ARM, but there's some kind of "Win10 IoT edition" thing I know very little about. |
| 00:09 | < RchrdB> | so now I'm wondering if it's possible to get something which, I guess ideally I'd like a tiny computer with an Atom or a high-end ARM in it? |
| 00:09 | <&McMartin> | Ah, I see. |
| 00:10 | <&McMartin> | The Surface Pros straight-up have Cores. |
| 00:10 | < RchrdB> | something with a 10W or less TDP that can go fanless, which AFAIK is one of the things that the Surface did |
| 00:10 | <&McMartin> | But it seems like this product line has diversified since I last looked at it. |
| 00:10 | <&McMartin> | And I don't know if the Pros can actually go fanless, since, as noted, they've got Cores~ |
| 00:11 | < RchrdB> | AFAIK they've got super low-voltage low-frequency versions with ~5W TDPs |
| 00:11 | <&McMartin> | Hmm. Looks like. |
| 00:12 | <&McMartin> | "m3-7Y30: 1.0 up to 2.6GHz, 4 MB cache, 4.5 W" |
| 00:12 | <&McMartin> | I don't know about the Core m-series, tbh |
| 00:12 | <~Vornicus> | I love how everybody else is talking about the power and I'm all "i cannot without pgdn" |
| 00:12 | <&McMartin> | I think the SP3 I had tested back in '14 was an i5. |
| 00:12 | <~Vornicus> | And I'm not even a command line jockey |
| 00:13 | <&McMartin> | I don't use PgDn on commandlines much. |
| 00:13 | <&McMartin> | But that wasn't his question! |
| 00:13 | < RchrdB> | according to ark.intel.com, the m3-whatever runs at about 5W and the low-power i5 runs at about 15W. |
| 00:13 | <~Vornicus> | well no, but |
| 00:13 | <&McMartin> | That is matching the specs on the Wiki page. |
| 00:13 | <&McMartin> | So it sounds like if you did rock this, you would definitely need the m3 version |
| 00:14 | <&McMartin> | Surface Pro 4, m3 version would be the closest in that line to what you want, but I've never run development environments on that chip with any OS |
| 00:15 | <&McMartin> | Aha! |
| 00:15 | <&McMartin> | Also, the m3 model is fanless, and the i-series ones are not. |
| 00:15 | <&McMartin> | How hardcore development are we talking here? |
| 00:16 | <&McMartin> | As long as the project size is measured in KLOC instead of MLOC I don't see how 900Mhz wouldn't be basically enough |
| 00:16 | <&McMartin> | (The project from 2014 in question was about 2 MLOC, because Qt and Boost) |
| 00:20 | < RchrdB> | well… don't know |
| 00:20 | <&McMartin> | I have, for about a decade now, preferred to do my heavy dev on i7-based machines, but any i7 CPU will fail your basic requirements. |
| 00:21 | < RchrdB> | I remember using a laptop with 32MB of RAM and I think a Pentium1 to learn computer science back in uni :) |
| 00:21 | <&McMartin> | Man, that was my *desktop* back in Uni. 120MHz. |
| 00:21 | < RchrdB> | alas, everything got big since then |
| 00:22 | <&McMartin> | I ran Borland Turbo C++ and Allegro Common Lisp on it. |
| 00:22 | <&McMartin> | The former was a DOS program, but that's OK, because Win95. |
| 00:23 | < RchrdB> | you probably still could under dosbox ;) |
| 00:23 | <&McMartin> | Can and have. |
| 00:23 | < RchrdB> | that reminds me |
| 00:23 | <&McMartin> | https://bumbershootsoft.wordpress.com/category/retrocoding/dos/ |
| 00:24 | <&McMartin> | Technically that's Turbo C, because that's the one that's the free download now. |
| 00:34 | < Jessikat> | McMartin: this one was more of a 'how you do this thing' than a 'why it works' |
| 00:34 | | * McMartin nods |
| 00:34 | < Jessikat> | the unimplemented template is so that you can write thing<one_template_param> |
| 00:34 | < Jessikat> | but the specialisation is the thing that's implemented, which matches on return_type(params...) |
| 00:34 | < Jessikat> | and gives names to them by the way it's formulated, so you can use them for operator(_) |
| 00:35 | < Jessikat> | oeprator()* |
| 00:35 | < RchrdB> | gah macbooks are expensive |
| 00:35 | < Jessikat> | some of these blog posts are going to be so I don't forget the things in future x) |
| 00:39 | | * TheWatcher does a first pass at documentation for Killbot, decides it is time to qwertyface |
| 00:40 | <@TheWatcher> | (TODO item #2: world domination) |
| 00:45 | < RchrdB> | christ a midrange android tablet would beat this laptop for actual usage right now |
| 00:45 | < RchrdB> | this thing is wrecked |
| 00:52 | <&McMartin> | ... killbot? |
| 00:53 | <&McMartin> | And yes, never pay the Mac Tax except under duress |
| 00:53 | <&McMartin> | You can *always* get more for less, unless the specific thing you need is OSX or applications that run only on it |
| 00:53 | <&McMartin> | ... which is basically XCode |
| 00:53 | < RchrdB> | but they're so shiny |
| 00:54 | < RchrdB> | and everything else has shit physical build quality (not that theirs is perfect either) |
| 00:55 | < RchrdB> | and 99% of everything else comes with Windows 10 ☹ |
| 00:56 | < RchrdB> | I only know of one place in the UK that sells laptops which don't have mandatory Windows on them |
| 00:56 | < RchrdB> | other than Apple I mean |
| 01:00 | <&McMartin> | Today in Programs That Didn't Work: https://i1.wp.com/bumbershootsoft.files.wordpress.com/2017/07/riscos_16_not_colo rbars.png |
| 01:01 | <&McMartin> | (Expected behavior: https://i0.wp.com/bumbershootsoft.files.wordpress.com/2017/07/riscos_15_colorbar s.png ) |
| 01:03 | <~Vornicus> | whups |
| 01:55 | | RchrdB [RchrdB@Nightstar-qe9.aug.187.81.IP] has quit [Ping timeout: 121 seconds] |
| 02:34 | | Jessikat [Jessica@Nightstar-bt5k4h.81.in-addr.arpa] has quit [[NS] Quit: Leaving] |
| 04:46 | <&[R]> | Okay, I'm compiling some old code I used to work on, it seems C has changed a bit since I last compiled it |
| 04:46 | <&McMartin> | How old is this code |
| 04:46 | <&[R]> | One of the things it does, is pass a pointer to a struct via va_args |
| 04:46 | <&[R]> | Less than 10 years, but close to that |
| 04:46 | <&[R]> | This struct has a union inside of it |
| 04:46 | <&McMartin> | Are you now compiling 64-bit instead of 32 |
| 04:47 | <&[R]> | Attempting to access the union results in a segfault |
| 04:47 | <&[R]> | Yes |
| 04:47 | <&McMartin> | Are you sure you're consuming the entire pointer |
| 04:47 | <&McMartin> | (Try logging the pointer values you are passing in and the ones you are receiving.) |
| 04:47 | <&[R]> | I have a printf that gives me the mem address before sending it and on recieving it |
| 04:48 | <&[R]> | They match up |
| 04:48 | <&McMartin> | How about the address of the union |
| 04:48 | <&McMartin> | Like, this should in fact work |
| 04:48 | <&[R]> | (gdb) print nd |
| 04:48 | <&[R]> | $4 = (struct npc_data *) 0x7ffffff7 |
| 04:48 | <&[R]> | (gdb) print nd->u |
| 04:48 | <&[R]> | Cannot access memory at address 0x8000005f |
| 04:48 | <&[R]> | If that helps any |
| 04:49 | <&[R]> | Note: this is my first time ever using gdb or any debugger |
| 04:49 | <&McMartin> | What's offsetof(npc_data, u)? |
| 04:49 | <&McMartin> | Also, um |
| 04:49 | <&McMartin> | Isn't it supposed to be nd->u.something? |
| 04:50 | <&McMartin> | offsetof(u) here looks like it would be 0x60, and you are crossing a page boundary there |
| 04:50 | <&McMartin> | Which is kind of unusual, ime |
| 04:50 | <&[R]> | (gdb) print nd->u.scr |
| 04:50 | <&[R]> | Cannot access memory at address 0x8000005f |
| 04:50 | <&McMartin> | k |
| 04:50 | <&[R]> | Program received signal SIGSEGV, Segmentation fault. |
| 04:50 | <&[R]> | 0x00000000004955de in npc_convertlabel_db (key=0x531657d, data=0x26, ap=0x7fffffffc1f8) at /root/tmw/eathena/src/map/npc.c:1308 |
| 04:50 | <&[R]> | 1308lst=nd->u.scr.label_list; |
| 04:50 | <&[R]> | Also VM |
| 04:51 | <&McMartin> | The address of nd looks kind of suspect to me, tbh |
| 04:53 | <&McMartin> | But I'd see if I couldn't put together a small example of this to try to repro it. |
| 04:54 | <&McMartin> | As long as both caller and callee were compiled at the same time, and you aren't linking against ancient objects, this "should" be fine. |
| 04:54 | <&McMartin> | But I bet the main issue is that this code isn't 32-bit clean somewhere else. |
| 04:54 | <&McMartin> | Can you compile with the -m32 flag across the board? |
| 04:54 | <&[R]> | New run: 000000007FFFFFEF is the output of printf("%016X", nd) before sending it. |
| 04:54 | <&[R]> | (gdb) print nd |
| 04:54 | <&[R]> | $1 = (struct npc_data *) 0x7fffffef |
| 04:55 | <&[R]> | Can try |
| 04:55 | <&McMartin> | Before you do though |
| 04:55 | <&McMartin> | Try accessing that union's valuse *before* the call |
| 04:55 | <&McMartin> | If it breaks before the call, it's not va_args's fault, for sure. |
| 04:56 | <&[R]> | I can, it actually gets filled a few instructions prior though |
| 04:56 | <&[R]> | I'm missing headers and such for compiling 32bit, and my distro's website is down |
| 04:56 | <&McMartin> | Boo. |
| 04:57 | <&McMartin> | Is this open-source/published somewhere? |
| 04:59 | <&McMartin> | Try checking the value of that union's members just before calling. It sounds a whole lot like something's scribbling over bits of stack it shouldn't. |
| 05:00 | <&McMartin> | Or straight-up freeing memory that's in use with something like brk(), but that seems vanishingly unlikely. |
| 05:00 | <&[R]> | It's open source yeah, I can upload the code for you, one sec |
| 05:00 | <&[R]> | Though one thing that's confusing me right now, is one line of my debugging code isn't being called on the segfaulting run |
| 05:01 | <&McMartin> | Really not being called or not getting a chance to be flushed out of stdout's buffers? |
| 05:01 | <&[R]> | 00000000052B7E18 - 00000000052B7DB0: 00000000052B7DB0 |
| 05:01 | <&[R]> | 000000007FFFFFDC |
| 05:01 | <&[R]> | Program received signal SIGSEGV, Segmentation fault. |
| 05:01 | <&[R]> | Two lines are output... |
| 05:01 | <&[R]> | // This is the calling environment |
| 05:01 | <&[R]> | printf("%016X - %016X: ", &(nd->u), nd); |
| 05:01 | <&[R]> | strdb_foreach(label_db,npc_convertlabel_db,nd); |
| 05:02 | <&[R]> | // This is inside the function where it segfaults |
| 05:02 | <&[R]> | printf("%016X\n", nd); |
| 05:02 | <&[R]> | fflush(stdout); |
| 05:02 | <&[R]> | npc_convertlabel_db is the function that segfaults |
| 05:03 | <&[R]> | http://rpgb.nobl.ca/npc.c.txt |
| 05:03 | <&[R]> | #0 0x00000000004955ee in npc_convertlabel_db (key=0x531657d, data=0x26, ap=0x7fffffffc1f8) at /root/tmw/eathena/src/map/npc.c:1309 |
| 05:03 | <&[R]> | #1 0x00000000004f47cc in db_foreach (table=0x52a7120, func=0x4954f1 <npc_convertlabel_db>) at /root/tmw/eathena/src/core/db.c:443 |
| 05:03 | <&[R]> | #2 0x0000000000496077 in npc_parse_script (w1=0x7fffffffcac0 "001-1.gat,0,0,0", w2=0x7fffffffcec0 "script", w3=0x7fffffffd2c0 "Mob001-1", w4=0x7fffffffd6c0 "-1,{", first_line=0x7fffffffdef0 "-1,{\n", fp=0x526afb0, lines=0x7fffffffdecc) |
| 05:03 | <&[R]> | at /root/tmw/eathena/src/map/npc.c:1507 |
| 05:03 | <&[R]> | #3 0x0000000000497f65 in do_init_npc () at /root/tmw/eathena/src/map/npc.c:2069 |
| 05:03 | <&[R]> | #4 0x0000000000482995 in do_init (argc=1, argv=0x7fffffffe478) at /root/tmw/eathena/src/map/map.c:1945 |
| 05:03 | <&[R]> | #5 0x00000000004f36c6 in main (argc=1, argv=0x7fffffffe478) at /root/tmw/eathena/src/core/core.c:123 |
| 05:03 | <&[R]> | Stack trace for line numbers |
| 05:05 | <&[R]> | Sending you the full source tree is a little more involved |
| 05:05 | <&McMartin> | I cannot help but notice that the strings "va_start" and "va_end" appear nowhere in this listing. |
| 05:06 | <&McMartin> | Those might have been no-ops in your old compiler, and perhaps they no longer are. |
| 05:06 | <&[R]> | I think that might be in db_foreach |
| 05:07 | <&McMartin> | I am looking specifically at npc_enable_sub. |
| 05:07 | <&[R]> | Actually, they'd have to be in the given functions wouldn't there? |
| 05:07 | <&McMartin> | Unless nullpo_retr is a macro and calls it, which seems unlikely given its other usages... |
| 05:08 | <&[R]> | nullpo_retr checks for null and will abort the function with a message to console. |
| 05:42 | | Turaiel[Offline] is now known as Turaiel |
| 05:48 | <&McMartin> | Bringing the conversation back here: Looks like the issue is "code isn't 32-bit clean" |
| 06:00 | <&[R]> | BTW, is doing stuff like: typedef struct {} Name; bad form at all in C? IIRC somepeople didn't like it, but I forget their rant |
| 06:01 | <&McMartin> | That's always been standard for me |
| 06:01 | <&McMartin> | Exactly as written, empty structs are iffy, and nameless structs can't be made recursive, which is sometimes a bummer. |
| 06:06 | <&[R]> | I meant as non-empty |
| 06:06 | | Turaiel is now known as Turaiel[Offline] |
| 06:06 | <&[R]> | But good to know |
| 06:07 | <&[R]> | Actually, question, why does gdb automatically have a break point on core.c:103? |
| 06:09 | <&McMartin> | Not offhand. That sounds odd. |
| 06:10 | <&[R]> | It's the first non-variable declaration line in main() if that matters any |
| 06:10 | <&McMartin> | Oh. |
| 06:10 | <&McMartin> | The program doesn't start automatically. |
| 06:12 | <&[R]> | That too |
| 06:12 | <&[R]> | But I have to do a "start" then a "continue" because of the automatic breakpoint |
| 06:12 | <&McMartin> | OK, I did not see that in my build |
| 06:12 | <&McMartin> | I just did "run" |
| 06:13 | <&[R]> | Weirdness |
| 06:18 | <&McMartin> | Aha |
| 06:18 | <&McMartin> | "start" and "run" are not the same thing |
| 06:18 | <&McMartin> | Start is "begin execution and stop immediately" |
| 06:18 | <&[R]> | Ah |
| 06:21 | <&[R]> | Is there a way to say in code "have a breakpoint here for a debugger"? |
| 06:22 | <&[R]> | ... actually I should probably just look up a tutorial |
| 06:31 | | celticminstrel [celticminst@Nightstar-cn2je9.dsl.bell.ca] has quit [[NS] Quit: And lo! The computer falls into a deep sleep, to awake again some other day!] |
| 07:07 | | Jessikat [Jessikat@Nightstar-0fcdb4.dab.02.net] has joined #code |
| 07:37 | | Kindamoody[zZz] is now known as Kindamoody |
| 08:14 | <&McMartin> | https://bumbershootsoft.wordpress.com/2017/07/09/project-mehitabel-graphics/ |
| 09:11 | | Jessikat` [Jessikat@Nightstar-o8kjr4.dab.02.net] has joined #code |
| 09:14 | | Jessikat [Jessikat@Nightstar-0fcdb4.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 11:09 | | Kindamoody is now known as Kindamoody|afk |
| 11:13 | | Jessikat [Jessica@Nightstar-bt5k4h.81.in-addr.arpa] has joined #code |
| 15:12 | | Jessikat [Jessica@Nightstar-bt5k4h.81.in-addr.arpa] has quit [Ping timeout: 121 seconds] |
| 15:40 | | celticminstrel [celticminst@Nightstar-cn2je9.dsl.bell.ca] has joined #code |
| 15:40 | | mode/#code [+o celticminstrel] by ChanServ |
| 15:41 | | Jessikat [Jessikat@Nightstar-kp04tl.dab.02.net] has joined #code |
| 15:44 | | Jessikat` [Jessikat@Nightstar-o8kjr4.dab.02.net] has quit [Ping timeout: 121 seconds] |
| 16:33 | | Jessikat [Jessikat@Nightstar-kp04tl.dab.02.net] has quit [[NS] Quit: Bye] |
| 17:08 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 17:30 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has joined #code |
| 17:30 | | mode/#code [+qo Vornicus Vornicus] by ChanServ |
| 17:44 | | Vornicus [Vorn@ServerAdministrator.Nightstar.Net] has quit [Ping timeout: 121 seconds] |
| 18:02 | | Jessikat [Jessica@Nightstar-bt5k4h.81.in-addr.arpa] has joined #code |
| 18:09 | | * [R] glares at gcc |
| 18:09 | <&[R]> | Thank you for erroring on your freaking massively long compile |
| 18:10 | <&[R]> | Because you were looking for cpp in /lib/ |
| 18:10 | <&[R]> | WHY ARE YOU FUCKING LOOKIGN THERE? |
| 18:19 | | Alek [Alek@Nightstar-7or629.il.comcast.net] has quit [Ping timeout: 121 seconds] |
| 18:23 | | Alek [Alek@Nightstar-7or629.il.comcast.net] has joined #code |
| 18:23 | | mode/#code [+o Alek] by ChanServ |
| 19:03 | | Derakon [Derakon@Nightstar-fr5qel.ca.comcast.net] has joined #code |
| 19:03 | | mode/#code [+ao Derakon Derakon] by ChanServ |
| 19:03 | <&Derakon> | "I put my pants on through lazily evaluated iterative deepening. I don't actually have any pants on until someone would notice I'm not wearing any." |
| 21:30 | < Jessikat> | XD |
| 21:32 | <&[R]> | Jessikat: how'd it go with your reading floppy mission? |
| 21:36 | < Jessikat> | oh! about 2/3 of them are readable, the most interesting thing I found was a firefox bookmarks list circa 2002 |
| 21:38 | <&[R]> | Nice |
| 22:45 | < Jessikat> | oh, and a zip file named sonic2.zip that I've not looked inside yet |
| 22:45 | < Jessikat> | for fear of what I might find |
| 23:17 | <@celticminstrel> | Heh. |
| 23:17 | <@celticminstrel> | At both. |
| 23:28 | | Kindamoody|afk is now known as Kindamoody |
| 23:52 | <@Azash> | [R]: The pleasure of an install dying because it had -Werror and an unused variable (!) In There Somewhere(tm) |
| 23:52 | <&[R]> | Heh |
| 23:52 | <&[R]> | Seriously though |
| 23:52 | <&[R]> | I've been trying to compile GCC for almost 20 hours now |
| 23:53 | <&McMartin> | gcc is very finicky about which version of gcc you use to compile it. |
| 23:53 | <&[R]> | 32bit one failed, but I think I need a newer kernel and relevant headers for that. |
| 23:54 | <@Azash> | Oh dear |
| 23:54 | <&[R]> | I'm noticing |
| 23:54 | <@Azash> | I thought you were just griping normally but |
| 23:54 | <&McMartin> | As I learned talking to the dude next to me on the plane last week: this was an explicit decision by the gcc steering committee |
| 23:54 | | Kindamoody is now known as Kindamoody[zZz] |
| 23:54 | <&McMartin> | They had tired of insisting on working on All The Compilers |
| 23:54 | <&[R]> | Seriously? |
| 23:54 | <&[R]> | WHY? |
| 23:55 | <&McMartin> | Because in 2000 they still had to use K&R style and avoid function prototypes because not all compilers accepted them |
| 23:55 | <&[R]> | I don't want to have to chain through 50 versions of GCC just to update an old install |
| 23:55 | <&McMartin> | And they were tired of that and felt it stalled continuing development. |
| 23:55 | | starkruzr [quassel@Nightstar-9848jp.fios.verizon.net] has quit [Ping timeout: 121 seconds] |
| 23:57 | <&McMartin> | It's more open than that |
| 23:57 | <&[R]> | Oh, so it's not "specific versions of GCC are required" it's "we need compilers that are at least slightly modern now" |
| 23:57 | <&McMartin> | It's in between that. |
| 23:58 | <&McMartin> | The thing I had complained about that set this off was that there was an SDK I was using that packaged its own GCC |
| 23:58 | <&McMartin> | And the compiler in Ubuntu 12.04 worked, and the compiler Ubuntu 16.04 did not. |
| 23:58 | | * [R] is going 4.5.0 to 5.4.0 |
| 23:58 | <&McMartin> | I was put out by this because my desktop computer is 8 years old |
| 23:58 | <&McMartin> | And a 4 year difference seems minor |
| 23:59 | <&McMartin> | But "it's based on Debian" does funny things to that. |
| --- Log closed Mon Jul 10 00:00:16 2017 |