Luanti logo

IRC log for #minetest-dev, 2014-02-20

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

All times shown according to UTC.

Enable nick filtering
Time Nick Message
10 more elements. Show/hide.
02:44 ShadowNinja ~tell sapier It doesn't work at all without cout/cerr.
02:44 ShadowBot ShadowNinja: O.K.
02:49 VanessaE ShadowNinja: regarding http://ix.io/atr , I think set the max length to 256
02:50 VanessaE seems that even 512 is enough to "fill" some folks' screens.
02:51 Exio4 but, wait, getting a message lost there, that is sad when you had to write 1025 characters to just ending in nothing
02:52 VanessaE so then maybe the client should retain it in the input buffer/dialog in such a case?
02:53 VanessaE if the client doesn't filter it out and it makes it to the server, well the server has little choice but to drop it, no?
02:53 Exio4 i would do it like minecraft (don't kill me, hey!), more than MAX_CHAT_LENGTH means no char is typed locally
02:53 VanessaE or the client should enforce a limit on the input length itself?
02:53 VanessaE yes, that
02:53 Exio4 what about cutting the string to the limit
02:54 domtron joined #minetest-dev
02:54 Exio4 cutting the string to the limit
02:54 Exio4 super-english!
5 more elements. Show/hide.
05:28 daspork I think my patch broke someting :P
05:28 daspork I need to play with a clean master to see if i can reproduce it... I have fallen through the world twice...
05:29 daspork I wasnt close to any of my custom nodes at the time so who knows lol
05:32 ShadowNinja Exio4: That's a unrealistically long line.  That would need to be split into at least three messages on IRC.
10 more elements. Show/hide.
11:59 Exio4 couldn't it split it too? if length>1024 then cut there; split first 256 chars; send those, call the same function with the rest of the string?
2 more elements. Show/hide.
12:22 celeron55_ 04:45:23 <+ShadowNinja> ~tell sapier It doesn't work at all without  cout/cerr.
12:22 celeron55_ wat
12:23 Jordach joined #minetest-dev
12:23 celeron55_ i think you simply fail to use minetest's log levels
12:23 celeron55_ how about learning how they work instead of assuming it's broken and using a stupid workaround
12:25 celeron55_ and dropping chat messages is stupid and will make anyone mad that writes a long enough message to get dropped
12:26 celeron55_ just make the client show only the beginning of long chat messages and show them fully in the console
12:26 celeron55_ or something sane like that
12:26 celeron55_ long messages aren't the server's problem
12:27 celeron55_ and why do i have to tell these to you, they're not hard to imagine yourself
12:32 PenguinDad joined #minetest-dev
12:36 CiaranG Ok, I think I've found a bug that's been bugging me for a long time. It goes like this:
12:37 CiaranG 1. map block becomes inactive, 2. 29 seconds (by default) later, the server async thread unloads it, 3. at the same time (within a second or two) the block becomes active again
12:38 CiaranG (actually, 2 and 3 happen in the opposite order)
12:40 CiaranG In other words, when a block becomes active, aside from making it load, nothing actually updates the 'last used' time, so there is still a short window when it can be unloaded
12:41 CiaranG This results in all manner of strange behaviour (from the player's perspective, placing blocks and having them disappear, occasionally, randomly, etc)
12:41 CiaranG Anyone able to say "yes, that sounds right" or "no, that's impossible"?
12:43 CiaranG I think I'm right, because I'm able to reproduce this quite regularly, but after inserting block->resetUsageTimer() at the top of ServerEnvironment::activateBlock it doesn't happen any more
28 more elements. Show/hide.
15:51 ShadowNinja celeron55_: I tried actionstream first, it interprets \r the same as \n.  And there's a reason for that drop.  I'll change it to a clip if it's so requested though.
15:56 iqualfragile joined #minetest-dev
15:59 sfan5 ShadowNinja: why would you want to use \r ?
2 more elements. Show/hide.
16:01 ShadowNinja sfan5: Because it rewrites the line rather than creating a new one.
16:02 sfan5 ik that, but for which specific case do you want to use this?
16:03 ShadowNinja sfan5: See the patch.  It makes backend migration less floody.
16:03 sfan5 ah
16:35 xyz that issue is so important!
16:36 sfan5 shall we maybe start to educate users (in the world creating menu) about map backends?
16:39 xyz no you should just switch to leveldb
16:43 xyz it's available on all platforms, isn't it?
16:44 xyz it works better
16:44 xyz so there's no reason to stay on sqlite
16:44 VanessaE minetestmapper.
16:44 sfan5 pff
16:44 sfan5 there are leveldb bindings for python
16:44 VanessaE less'n you're volunteering to rewrite both the C++ and python ones, that is.
16:44 sfan5 that's no reason
16:45 sfan5 you need to change some stuff
16:45 sfan5 but not much
16:45 xyz and it's shit anyway
16:45 VanessaE sfan5:  who honestly still uses that slow-ass python one?
16:45 sfan5 VanessaE: dunno
16:45 VanessaE the C++ one is like 50 times faster
16:45 xyz is it in main source tree?
16:45 sfan5 also sqlite3 is a database and we use it for key-value storage
16:46 sfan5 has anyone taken a look at rocksdb?
16:46 VanessaE xyz:  I've asked for it to be included.  no one gave a good reason why it can't be.
16:46 xyz ah
16:46 xyz that sucks
16:46 VanessaE https://github.com/ShadowNinja/minetest-mapper-cpp
16:46 VanessaE ShadowNinja's fork has the fewest bugs :)
16:46 xyz sfan5: there's literally no reason to change databases randomly
16:46 xyz what we see is: sqlite is slow, leveldb is fast
16:46 VanessaE (the only bug I know of actually is that the map/server needs to be completely idle or it'll produce a broken image)
16:47 xyz fast enough so you won't notice a difference by switching to another kv storage
16:47 xyz and sqlite is slow enough to notice the difference
16:47 xyz takes several seconds to flush iir
16:47 xyz iirc*
16:48 xyz VanessaE: maybe someone should incorporate it into main source tree, so it won't need to read map by itself
16:48 xyz and just use available map backends
16:48 VanessaE xyz:  it would be nice.
16:48 VanessaE that util is WAY too handy to not have around
16:49 VanessaE and it's more versatile than the python one as well
16:49 VanessaE it can create much larger images, for example
16:49 xyz but still no isometric
16:49 VanessaE yeah
16:49 xyz which is unfortunate
16:50 VanessaE onomatopoeia was nice looking
16:51 xyz ah right
16:52 xyz but then I found out that developing an isometric mapper isn't fun
4 more elements. Show/hide.
17:10 kahrl being a util, the C++ mapper falls into startup/config/util, right? so I guess I will pronounce that it shall be included in the main tree
17:11 VanessaE \o/
2 more elements. Show/hide.
17:17 VanessaE oh I forgot to mention..  xyz, PenguinDad mentioned earlier that he couldn't connect to one of my servers using freeminer.  He did not indicate what version/git commit.
17:20 VanessaE probably related to my using sapier's pull (1143)
17:20 NakedFury joined #minetest-dev
17:21 proller 99% its from sapiers connection remi
17:21 proller x
17:32 xyz VanessaE: then ignore him until he says what version he used
17:33 VanessaE I didn't have a chance to reply, only saw it in my scrollback.
17:33 PengunDad I tried both the "stable" Version of freeminer and latest commit on master
17:33 VanessaE hah, there he is :P
17:35 xyz well then it's kinda bad :(
3 more elements. Show/hide.
18:27 Selat Hello. What about this pull request https://github.com/minetest/minetest/pull/1151?
18:30 PilzAdam Selat, its not good that you removed 2 algorithms (juding by your changes to the docs)
18:32 proller who use this algorithms ?
18:33 Selat I already written that there were only two algorithms. One of them was shit thqat shouldn't used in such case.
18:33 celeron55_ who made the original code? sapier?
18:34 PilzAdam celeron55_, yes
18:34 PilzAdam its only used in mobf, but there might be other mods that use it
18:35 celeron55_ i like the diff stats on that pull request, assuming it doesn't remove anything that was actually useful and doesn't have huge speed regressions
18:36 Selat It have huge spped improvement in average case.
18:36 proller original was slower than lua in some cases  (by sapier tests)
2 more elements. Show/hide.
18:47 xyz Selat: (offtopic) do you have any background in programming contests?
18:51 VanessaE inb4 mr. 'all numbers' username on the forums constant requests for mods with deadlines of late in the year is mentioned. :)
18:53 Selat yes
18:54 xyz I see; expected that
18:59 xyz VanessaE: what was this about?
19:00 sfan5 xyz: have a diagram: http://i.imgur.com/9S5K0TV.png
19:01 VanessaE xyz:  when you mentioned programming contests, the first thing I thought of was that guy who was trolling the forums.  ALL CAPS posts, thinks he's "all that", all-numeric username, trying to get others to code MC-alike mods for him, promising some kind of vague prize to be awarded late in the year, etc.
19:03 xyz wait, how's that related to programming contests?
19:03 xyz sfan5: that diagram should be in caps
19:03 sfan5 haha
19:05 rubenwardy That username sucks. Someone could easily make an account that looks like him. Not that they would want to.
19:06 farlepet joined #minetest-dev
19:07 VanessaE xyz: because the way his posts are written, that's exactly what he acts like he's trying to accomplish?
19:07 VanessaE a contest, as in a competition to produce the best of something?
19:08 sfan5 I'm sure he/she will claim the price and release it as his/her mod
19:14 VanessaE probably :P
19:24 xyz I actually referred to things like ACM http://en.wikipedia.org/wiki/ACM_International_Collegiate_Programming_Contest
197 more elements. Show/hide.

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