Minetest logo

IRC log for #minetest-dev, 2014-05-28

| Channels | #minetest-dev index | Today | | Google Search | Plaintext

All times shown according to UTC.

Time Nick Message
00:19 MichaelRpdx joined #minetest-dev
00:26 kaeza can the items in builtin/game/deprecated.lua be removed?
00:26 kaeza specifically, digprop_*
00:26 kaeza also node_metadata_inventory_move_allow_all
00:29 kaeza also, isn't it about time .env: dies?
00:29 kaeza meh... too many mods still use it
00:32 SoniEx2 joined #minetest-dev
00:41 Jordach joined #minetest-dev
01:09 ShadowNinja kaeza: A lot if that can probably be removed by now.  Env: is still used a lot though, and there may be others.
01:36 kaeza https://github.com/kaeza/minetest/commit/​f3a759ecb91b5f786c20fad180e0d9cb0aa003d4
01:36 kaeza ShadowNinja, ^
01:37 LemonLake :D
01:37 LemonLake You took my suggsetion!
01:41 LemonLake <LemonLake> In 0.4.10 there should be a major cleanup
01:51 NakedFury joined #minetest-dev
01:56 BrandonReese joined #minetest-dev
01:57 kaeza bleh, this one instead https://github.com/kaeza/minetest/commit/​d57f54fa27380d69740f972f343ba410dece192c
01:58 LemonLake It's the same?
01:58 kaeza (forgot to commit the removal of deprecated stuff)
01:58 LemonLake Ah I see
01:58 LemonLake I'd like to see how many mods that breaks :P
01:58 kaeza test it!
01:59 VanessaE define "deprecated stuff"......
01:59 LemonLake I'm working on my own thing atm
01:59 kaeza VanessaE, see above
01:59 LemonLake Removal of aliases for core.register_alias
01:59 kaeza digprop_* is gone! :D
01:59 kaeza are*
01:59 VanessaE oh yeah
01:59 LemonLake Use of the old property system
01:59 LemonLake Digprop
02:00 VanessaE just so long as you don't break anything that's reasonably current :)
02:00 VanessaE I don't have to fork and fix half the mods used in dreambuilder :P
02:00 kaeza I joined in Oct 2012 and it was deprecated back then
02:00 kaeza sooooooo...
02:00 kaeza blame the modders :P
02:01 VanessaE well I mean more like env.:*
02:01 VanessaE er .env:*
02:01 ShadowNinja Kaeza: Looks good except for:  1. Conflict with my chat command return values pull :-|    <del>2. Deprecated things can be removed.    </del>3. do_nodeupdate is pointless.
02:01 kaeza no, it's still there
02:02 LemonLake Guys, if I have a table with other tables in it, and I nil all the values in one table, how can I shift the other tables 'down'
02:02 LemonLake Sorta like move the values of 2-5 to 1-4
02:02 ShadowNinja LemonLake: Far from a major cleanup.  Minetest has been undergoing cleanups of many sizes for a while.
02:03 LemonLake Somewhat of my suggestion
02:03 LemonLake We were talking about the mess in builtin
02:03 ShadowNinja LemonLake: look at table.remove.
02:03 LemonLake Yes but I'd rather just shift the values down
02:04 kaeza remove == shift down
02:04 LemonLake This is a hud I'm editing here
02:04 LemonLake I have 5 'lines', I blank the values of one to 'remove it' and I want to shift the rest
02:04 ShadowNinja LemonLake: -> -mods
02:04 kaeza ^ that too
02:05 LemonLake Sapier told me it was unwise to be removing each hud
02:05 LemonLake and replacing it, so to just have 5 lines and change them as necessary
02:05 LemonLake '-mods'?
02:05 ShadowNinja LemonLake: #minetest-mods
02:05 kaeza #minetest-mods
02:05 kaeza ShadowNinja'd :P
02:05 LemonLake Just realised what channel this was rofl
02:31 VanessaE error discovered by cheapie:  migrating a world that has a bad block crashes the minetest instance being used to migrate.
02:32 VanessaE what "order" does migration take?  I mean, what's done first?  bottom-to-top I support?
02:32 VanessaE s/support/suppose/
02:33 LemonLake What's with this s/ thing? Is it part of a client or something?
02:33 hmmmm joined #minetest-dev
02:33 VanessaE (migrating sqlite --> leveldb)
02:33 VanessaE LemonLake: it's shorhand, from the unix 'sed' command
02:33 VanessaE it means to substitute the second string for the first one.
02:33 LemonLake Ah, just a more sense making alternative to just using an asterisk
02:34 ShadowNinja VanessaE: The order returned by getAllLoadableBlocks, which means no partcular order.
02:34 LemonLake ew
02:34 LemonLake but also un-ew
02:34 LemonLake ø.ø
02:34 VanessaE hrm
02:36 VanessaE ShadowNinja: cheapie is doing a migration test on Vanilla, which you may remember has a bad mapblock up close to 0,+31000,0.  when it hits that bad block (very close to the end of the process), it goes boom :)
02:37 VanessaE [05-27 22:28] <cheapie> 21:27:36: ACTION[main]: Migrated 6254500 blocks 99.9969% completed
02:37 VanessaE [05-27 22:28] [...] Invalid block data in database (-1,1937,0) (SerializationError): unsupported NameIdMapping version
02:37 VanessaE [05-27 22:28] <cheapie> 21:27:37: ERROR[main]: Ignoring block load error. Duck and cover! (ignore_world_load_errors)
02:37 VanessaE [05-27 22:28] <cheapie> Segmentation fault (core dumped)
02:37 LemonLake What precisely is a 'bad mapblock'?
02:37 LemonLake It looks near the limit
02:37 VanessaE LemonLake: a mapblock in the map file that has become corrupted, in this case by entirely unknown means.
02:37 LemonLake Dang
02:38 LemonLake Can't you entirely remove the mapblock from the database, regenerate it then run the migration process again?
02:39 VanessaE you can but damned if I know how.
02:39 VanessaE ShadowNinja: the above error can be reproduced - cheapie did it using the current downloadable copy of my Vanilla world.
02:39 VanessaE (it's a big file, but it's there for the taking)
02:39 LemonLake I can interpret databases fine enough, it just depends on how it's actually formatted
02:39 LemonLake Where can this be downloaded?
02:40 VanessaE http://digitalaudioconcepts.com/vanessa/hobbie​s/minetest/worlds/Vanilla_World-map.sqlite.bz2
02:40 LemonLake I'll have a look at the file when it's downloaded
02:41 LemonLake I assume it's compressed, how much compression do you know?
02:53 VanessaE zippo :)
02:53 VanessaE and I don't think the map is compressed
02:53 VanessaE (though the above download obviously is)
02:56 lemonlake_ joined #minetest-dev
02:57 lemonlake_ VanessaE: I tried, but halfway through loading the 1.4gb file the entire machine hanged
02:58 VanessaE ouch
02:58 VanessaE WOW
02:58 VanessaE who has the 100 Mbps pipe???
02:58 VanessaE holy shit
02:58 lemonlake_ The only thing I can do is move the mouse cursor, everything else is just dead
02:58 lemonlake_ But I gotta sleep, I'll try again tomorrow. Cya
02:59 * VanessaE just noticed the two bandwidth spikes :)
03:13 ShadowNinja VanessaE: I made a little C++ program to compute a block hash for me to nuke a bad block in my DB...
03:18 VanessaE well that's one way to do it
03:18 VanessaE won't fix the segfault though :)
03:18 VanessaE I'll try it later on and see if I can get a backtrace
03:34 ShadowNinja VanessaE: http://sprunge.us/hiSN
03:34 kaeza joined #minetest-dev
03:36 VanessaE I think I have a bash script here with a similar function
03:37 VanessaE hm, nope.  it just has the formula.
03:45 Amaz_ joined #minetest-dev
03:59 Amaz_ joined #minetest-dev
05:30 werwerwer_ joined #minetest-dev
05:44 Hunterz joined #minetest-dev
06:05 SoniEx2 joined #minetest-dev
06:11 PenguinDad joined #minetest-dev
06:44 proller joined #minetest-dev
06:45 emptty joined #minetest-dev
06:47 darkrose joined #minetest-dev
07:00 Garmine joined #minetest-dev
07:53 SmugLeaf joined #minetest-dev
07:55 grrk-bzzt joined #minetest-dev
08:13 rdococ joined #minetest-dev
08:52 LemonLake joined #minetest-dev
08:58 eeew joined #minetest-dev
09:06 lemonlake_ joined #minetest-dev
10:01 VanessaE Is there some particular reason the --migrate process ignores the --port switch and enforces using port 30003 ?
10:40 vifino joined #minetest-dev
10:43 deltib joined #minetest-dev
10:44 ImQ009 joined #minetest-dev
11:06 LemonLake joined #minetest-dev
11:24 kaeza joined #minetest-dev
11:32 Megaf joined #minetest-dev
11:47 Megaf Olá
11:47 Megaf that means, hello
11:53 grrk-bzzt joined #minetest-dev
11:56 vifino joined #minetest-dev
12:30 LemonLake joined #minetest-dev
12:45 Jordach joined #minetest-dev
13:07 dsimon coi
13:07 dsimon That does too :-)
13:10 NakedFury joined #minetest-dev
13:17 hmmmm joined #minetest-dev
13:45 grrk-bzzt joined #minetest-dev
13:55 smoke_fumus joined #minetest-dev
14:22 PilzAdam joined #minetest-dev
14:23 zat joined #minetest-dev
14:45 rubenwardy joined #minetest-dev
14:54 mrtux joined #minetest-dev
15:00 LemonLake joined #minetest-dev
15:11 Zeitgeist_ joined #minetest-dev
15:19 kaeza no comments on #1341 ? :(
15:19 ShadowBot https://github.com/minetest/minetest/issues/1341 -- Builtin cleanup, part 1 (game). by kaeza
15:20 LemonLake I gave you a comment yesterday
16:02 Calinou joined #minetest-dev
16:46 Garmine joined #minetest-dev
16:47 Hunterz joined #minetest-dev
16:52 sapier joined #minetest-dev
17:09 smoke_fumus Streaming gamedev right now http://www.twitch.tv/smoke_th
17:10 Calinou this is a serious channel
17:10 * Calinou shows smoke_fumus and Unity 3D the door
17:19 zat joined #minetest-dev
17:20 PenguinDad joined #minetest-dev
17:21 Piggybear87 joined #minetest-dev
17:25 restcoser joined #minetest-dev
17:27 smoke_fumus Calinou: yes and i am a serious developer
17:27 kaeza joined #minetest-dev
17:28 grrk-bzzt joined #minetest-dev
17:41 Megaf Fixes some variables being leaked to the global environment (Green Peace would be proud).
17:41 Megaf heh
17:44 Megaf well done kaeza
17:48 proller joined #minetest-dev
17:49 Piggybear87 joined #minetest-dev
18:28 LemonLake joined #minetest-dev
18:35 vifino joined #minetest-dev
18:41 LemonLake joined #minetest-dev
18:52 grrk-bzzt joined #minetest-dev
19:12 tomreyn joined #minetest-dev
19:54 grrk-bzzt joined #minetest-dev
20:16 pitriss joined #minetest-dev
20:42 kahrl joined #minetest-dev
21:19 Piggybear87 joined #minetest-dev
21:55 werwerwer joined #minetest-dev
22:01 SoniEx2 joined #minetest-dev
22:02 ShadowNinja VanessaE: The fact that it listens at all, or loads mods, should be considered a bug.  But the map depends on the server, or at least did, although I think kwolekr reduced or eliminated that dependency.
22:11 VanessaE well I don't mind its dependencies, some of that makes sense since it does call upon the server to do the hard work
22:11 VanessaE just that hard-coded port 30003 was all I was concerned about
22:12 VanessaE by "the server" of course I mean the engine core
22:13 VanessaE I just migrated all but Vanilla to leveldb, but I had to keep MG-Nore shutdown the whole time (even after it was done), because it uses port 30003 as does the migrate tool.
22:17 ShadowNinja sapier: Can you comment on #1129?
22:17 ShadowBot https://github.com/minetest/minetest/issues/1129 -- Add output and success return values to chatcommands by ShadowNinja
22:18 ShadowNinja sapier: I've switched the order back to (success, message) and done some cleanup locally.
22:19 sapier are you sure?
22:19 ShadowNinja Branch updated with local changes.
22:19 ShadowNinja sapier: Sure about what?
22:19 sapier return done and "Done." or "Failed.", done seems to be text first then bool value
22:20 ShadowNinja sapier: Check second commit.
22:21 sapier ahhh ... sorry it's late :-)
22:22 sapier -- Returns text output and boolean success. ... that has to be changed back too
22:25 ShadowNinja Oops, missed the docs.
22:25 sapier I'm positive about that fix but it's way more important to find a fix for the async problems, please don't forget about that one
22:27 ShadowNinja sapier: So, you approve it?
22:27 sapier I already commented this way yes ... I assume you're there to fix bugs caused by it ;-)
22:28 ShadowNinja O.K.  :-)  I'll do some final testing, squash, merge, and push.
22:28 sapier I don't expect it to cause major issues, but things like that look like a few minor issues
22:32 ImQ009 joined #minetest-dev
22:42 ShadowNinja Pushed and dev wiki updated.
22:42 ShadowNinja Oh, I updated the Changelog a day or two ago BTW.
22:44 sapier did you find other issues in #1200? I'd merge it tomorrow evening and prepare final step of merging android code afterwards
22:44 ShadowBot https://github.com/minetest/minetest/issues/1200 -- Add support for exiting formspecs by doubleclicking outside by sapier
22:48 ShadowNinja sapier: On line 2379 you can invert the conditional, have it return, and reduce the indentation level by one.
22:48 ShadowNinja Same for 2385.
22:49 sapier You're right I'm gonna change that too
22:49 LemonLake Oh btw sapier
22:49 LemonLake I got the death messages working
22:49 sapier congratulations
22:51 ShadowNinja sapier: Is getDeltaMs subbosed to return a absolute value, or does it return negatives too?  Because the code seems unnecessarily complex and seems to depend on the int size.
22:52 sapier delta is per definition always positive
22:52 sapier and no it doesn't depend on int size as all parameters are specified as 32bit unsigned values
22:52 ShadowNinja sapier: Wouldn't just "return abs(new - old);" work?
22:53 sapier that'd fail for 64 bit os
22:53 ShadowNinja sapier: I mean it depends on 32 bit ints.  You can't just change the parameters and return value to u64, for example.
22:53 sapier no it doesn't depend on it
22:53 sapier int
22:54 SoniEx2 joined #minetest-dev
22:54 sapier as int is always system with while u32 is specified
22:54 sapier we've got 32bit timestamps only
22:55 sapier I'd prefere not to use specific datatypes too but I don't intend to change any single time usage within minetest ;-)
22:55 ShadowNinja The logic depends on the size.
22:55 ShadowNinja sapier: How can a simple operation like abs or sibtraction depend in the archetecture?
22:56 sapier overflow
22:57 ShadowNinja sapier: The difference if only one bit.
22:57 ShadowNinja is*
22:57 ShadowNinja If it overflows with one less bit you should be using 64 bit ints.
22:57 sapier and we're doing unsigned operations so it's even worse
22:58 ShadowNinja sapier: And it doesn't belong it porting, it's simple math.
22:58 sapier well abs on unsigned doesn't make sense anyway
22:58 sapier unsigned values are always positive
22:59 ShadowNinja sapier: Yes, that's why it would get a signed value, or you'd use some custom logic for it that doesn't depent on the int size.
22:59 ShadowNinja (not referring to the C "int" type)
23:00 sapier this code doesn't depend on int size
23:01 ShadowNinja sapier: On 2751 you should probably use a size_t.
23:01 sapier the way it's written now it will work on 16bit 32bit 64bit as well as 128bit
23:01 ShadowNinja sapier: NOT referring to the C "int" type.
23:01 ShadowNinja I mean it in the mathematical sense.
23:01 sapier a indes isn't a  size so I don't agree to this
23:01 ShadowNinja If I changed the arguments to u64 it would fail.
23:02 sapier well if you change the arguments it's not the current code any longer ;-P
23:02 ShadowNinja I mean that the logic depends on the prototype.
23:02 sapier yes and it's intended to be this way
23:03 sapier this function is supposed to work for 32bit timestamps no more no less. You can suggest a generic code too if you know one
23:04 ShadowNinja sapier: Use something like this: v
23:04 ShadowNinja http://pastebin.ubuntu.com/7539902/
23:04 ShadowNinja And you can make that into a generic delta function or #define.
23:05 ShadowNinja Well, not a macro, due to the returns.
23:05 sapier maybe something but obviously not that code ;-)
23:05 ShadowNinja sapier: Hmmm?
23:05 VanessaE ShadowNinja: <offtopic> map safety question..
23:05 sapier well you'll get negative --> huge values using that code
23:06 ShadowNinja sapier: How?
23:06 sapier well if new > old then old -new is negative ... for unsigned a huge value
23:07 ShadowNinja sapier: Er, s/>/</ or course ;-)
23:07 ShadowNinja of*
23:07 sapier still the problem with system specific int size remains
23:08 sapier our timestamps are only 32 bit
23:08 ShadowNinja sapier: No.  Use a u32 there.
23:08 VanessaE um, strike that safety question.  I just got my answer :D
23:08 ShadowNinja sapier: The code just doesn't depend on the fact that we're using a u32, and it's much clearer.
23:09 ShadowNinja sapier: Other than that the only other issues I notice are code style issues throughout. (s/)\n{/) {/ and s/,/, /)
23:10 sapier to me my variant is much more clean :-) while your's is most likely equivalent
23:10 sapier still there? I thought I fixed all of them by now
23:11 VanessaE fwiw I was about to ask:  I disabled load errors and used worldedit to place a cobble block into the failed mapblock.  it threw a warning about a large number of objects in the map therein (about 28000) which it claimed to remove.  I exited the game and restarted.  the bad mapblock warning was gone but another 81k entities were allegedly there.  the engine eventually ate all available RAM + swap :)
23:11 sapier lol
23:12 sapier ok shadow I'm gonna change the delta fct and change the if logics but I wont use size_t for a index value
23:12 * VanessaE is now attempting to migrate the modified map to leveldb, just for shits and grins ;)
23:13 sapier why do you migrate to leveldb?
23:21 VanessaE it's faster
23:21 sapier really?
23:21 VanessaE and with the system I run these servers on I need all the speed I can get :)
23:21 sapier or is it only the sync issue?
23:22 VanessaE and the map files are smaller
23:22 sapier ok guess your server is a reasonable testbed to check if trading consitency for speed is a good idea ;-)
23:24 VanessaE my creative map dropped from 2.2 GB down to 1.3 GB
23:24 sapier interesting
23:24 VanessaE Survival went from 4.9 GB to 3.2 GB
23:25 VanessaE other maps saw similar changes.
23:25 sapier if you're test result is positive I suggestdiscussing about switching default db
23:25 VanessaE c55 wants that sync issue explored first
23:25 sapier that's what I mean with "positive"
23:35 VanessaE one thing's for sure, the worlds certainly load fast as hell
23:45 sapier left #minetest-dev
23:59 ImQ009 joined #minetest-dev

| Channels | #minetest-dev index | Today | | Google Search | Plaintext