#jsmess 2017-06-30,Fri

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)

WhoWhatWhen
JohnTalencloser.
https://pastebin.com/rMaNexTN
[00:05]
closer.
ps2mouse conflicting.
which means recompiling everything lol.
[00:13]
............ (idle for 55mn)
***bwn has quit IRC (Read error: Connection reset by peer) [01:08]
JohnTaleni could use a spare mind. I'm getting symbol multiply defined. But I checked the entire project with grep and it is extern everywhere but mouse.c. [01:14]
baiwhat's the symbol? [01:15]
JohnTalen_mouse_enabled_
getting much closer though.
i am at the last step. but it's all the substeps that I possibly have to go through.
bai: want me to send you a google drive location?
i've only used grep, but grep should definitely catch it.
[01:16]
***bwn has joined #jsmess [01:20]
JohnTaleni would only say that SDL could have _mouse_enabled. I'll have to check.
gotta take a break. argh.
i'ts 222megs. I kept the .o .i .bc intact.
just need an email.
bbiaf
[01:22]
baioh sorry, had to take care of some stuff for dinner. hmm. that sounds like a symbol that's defined in the app, rather than a library... [01:32]
JohnTalenyeah it is. [01:32]
baimost of the sdl functions are prefixed [01:32]
JohnTalenI had only thought SDL may define it. [01:32]
baiok
yeah
[01:32]
JohnTalenok great. [01:32]
baithat's what I was wondering as well
but shouldn't be the case
[01:32]
JohnTaleni can taste the js! lol.
i'll check emscripten then.
aparently not there either.
there was a 'int _mouse_enabled=0;' in the ps2mouse.c file but i put an extern in front of it. clang doesn't complain. but emcc is.
error: Linking globals named '_mouse_enabled': symbol multiply defined!
this shouldn't be rocket science. maybe it's just telling me tomorrow.
shit. for some reason it showed up in my target.
oh no. that's ps2mouse_enabled;
i'll have to fine comb tooth this tomorrow.
[01:33]
......... (idle for 44mn)
got past it. i removed the extern and just included the original in the ps2mouse.c definition. Now I have another one on mouse_resource_init. Slowly, but surely. [02:29]
baicool [02:34]
JohnTalenyah.
2017 and errors are a one bucket linked list. go figure.
[02:35]
closer. [02:49]
....................... (idle for 1h51mn)
Lord_Nighif there's any issues with jsmess which the MAME team needs to know about, they're on freenode #mame and #mame-dev
and a few devs are here too
[04:40]
............... (idle for 1h12mn)
***JohnTalen has quit IRC (Read error: Operation timed out) [05:52]
........................................... (idle for 3h30mn)
davidar has quit IRC (Quit: Connection closed for inactivity) [09:22]
......... (idle for 40mn)
davidar has joined #jsmess [10:02]
.................................................................................................... (idle for 8h17mn)
DopefishJ has joined #jsmess
DFJustin has quit IRC (Ping timeout: 260 seconds)
[18:19]
.................. (idle for 1h27mn)
JohnTalen has joined #jsmess [19:48]
JohnTalenAnyone familiar with this clang bug https://stackoverflow.com/questions/7276828/how-to-stop-clang-from-duplicating-functions-in-standard-c-header-files ?
It's big in vice and now I'm at a defintinition I can't 'wish away'. I tried -std gnu89 to no avail. Still get 'multiply applied'.
The goal is not have the change the whole project to stay gnu compliant.
[19:49]
My only attempt will be to upgrade to the latest clang. it's really a shot in the dark.
i am using clang 3.8.1-24
[19:57]
Vito`JohnTalen: did you see the last comment here: https://bugs.llvm.org/show_bug.cgi?id=5960#c10
but also it does look like in some platforms that VICE removes inlining
for example, --enable-inline in the configure script can be set to 'no'
and in macosx, which would use clang to compile, at least one library gets all its "inline"s sed'd out
lots of ifdefs about inlining also
[19:59]
JohnTalenVito`: great thanks. I found this: https://stackoverflow.com/questions/12747198/compiling-error-when-std-gnu99-and-inline-function-is-used
I will give these a try. Much appreciated.
[20:08]
........ (idle for 37mn)
enable inline can only be set to 'set' in this script.
and inlining is different than 'multiply specified'.
[20:45]
......... (idle for 43mn)
this is obviously a compiler bug. machine_class is not multiply defined. all the other bugs I fixed I more or less commented out because of non specification (like scpu and c64dtv).
so i tried '-std=gnu89 -fgnu89-online' and they do *not work*. I'd hate to attempt to compile clang from source since this clang release is only a year old.
[21:28]
i guess i'll try gnu_inlnie.
i guess i'll try gnu_inline.
Vito`: it says to add it the glibc errors. Is this suggesting to recompile glib.
glib is almost impossible to upgrade on linux systems. it's vital for so many things.
Vito`: ^ .=?
[21:35]
Vito`I assume you're not actually running an antique version of Linux with an antique glibc
My understanding of the problem is that inlining in some cases duplicates definitions
[21:40]
JohnTalenLinux la-flamouge 4.9.0-3-amd64 #1 SMP Debian 4.9.30-2+deb9u1 (2017-06-18) x86_64 GNU/Linux [21:41]
Vito`the last comment on that page suggests adding __attribute__ ((__gnu_inline__)) to whatever headers you're experiencing the problem with might change the semantics of inlining enough that it won't happen
also, if you're reading the configure file literally, it's not checking to be "set" it's checking whether it's set, as in, whether it's enabled, which is the default
[21:42]
JohnTalenok
i think the name confliction is the _class appendage. I'll just change the name. easy enough.
nah, 234 changes isn't worth it... yet
away she goes.
nope __attribute__ ((__gnu_inline__)) didn't work.
okay, i'm changing the name.
[21:46]
......... (idle for 40mn)
i had it compiling to .i and .bc before which is nice but not anymore. at least it's generating .o LLVM bitcode which is the ultimate goal.
i made the changes to the global variable name, compiling as I speak.
--emit-llvm breaks my configure. Vito` any ideas?
It's not critical, thought you'd might know.
i'm using save-temps which does work, but doesn't output the nice .i .bc files.
well, it's not the actual variable name. still throws up on it.
wow, this isn't easy.
compiling js...
error: Linking globals named 'machine_c64_variety': symbol multiply defined!
it is multiply defined but within reason.
i'll just force extern.
[22:41]
.... (idle for 17mn)
okay, i just defined it in one .h file and *that* is better. [23:03]
...... (idle for 26mn)
***Swizzle has joined #jsmess [23:29]
JohnTalenso bizarre, the multiple defined variable. it's like... old files. hm. [23:40]
i know it's not the temp files nor the variable declarations. i fixed. this. wtf. [23:46]

↑back Search ←Prev date Next date→ Show only urls(Click on time to select a line by its url)