Time |
Nick |
Message |
00:35 |
|
Weedy_lappy joined #minetest-dev |
02:03 |
|
ssieb joined #minetest-dev |
02:57 |
|
proller joined #minetest-dev |
03:53 |
|
proller joined #minetest-dev |
04:22 |
|
neko259 joined #minetest-dev |
05:04 |
|
smoke_fumus joined #minetest-dev |
06:03 |
|
Akien joined #minetest-dev |
06:04 |
|
kaeza joined #minetest-dev |
06:05 |
|
darkrose joined #minetest-dev |
06:05 |
|
darkrose joined #minetest-dev |
06:36 |
|
kaeza joined #minetest-dev |
07:08 |
|
kaeza joined #minetest-dev |
08:18 |
|
Calinou joined #minetest-dev |
09:03 |
celeron55 |
should we disallow using the minetest logo as an avatar on the forum? |
09:03 |
celeron55 |
or at least allow it only for core devs or something |
09:04 |
celeron55 |
many people have started using it and it's just worthless as an avatar now |
09:05 |
VanessaE |
I would tend to agree with that idea |
09:06 |
Calinou |
that's a silly rule |
09:06 |
kaeza |
Allow it, but allow some other special overlay for devs? |
09:06 |
kaeza |
like the MT icon plus an overlaid "DEV" or something |
09:07 |
* celeron55 |
does not like kaeza's suggestion |
09:08 |
kaeza |
ah well, some random suggestion :P |
09:10 |
* sfan5 |
has apples on the tree in the logo and a lava lake |
09:10 |
sfan5 |
s/in the logo and a lava lake/and a lava lake in the logo/ |
09:33 |
thexyz |
celeron55: we should disallow avatars |
09:33 |
thexyz |
and use something like gravatar |
09:33 |
thexyz |
just make it so users cannot change it |
09:34 |
VanessaE |
gravatar is okay, but not having avatars at all? what is this, 1990? |
09:34 |
thexyz |
no, this is 2013 |
09:35 |
VanessaE |
exactly. |
09:35 |
VanessaE |
there's no legit reason to get rid of them. |
09:35 |
thexyz |
no there is |
09:35 |
thexyz |
they distract people |
09:36 |
thexyz |
s/no/yes/ |
09:36 |
VanessaE |
if you're distracted by a little icon off to one side... um... |
09:37 |
Calinou |
avatars allow for quick identification |
09:37 |
Calinou |
and you can still change a gravatar |
09:38 |
thexyz |
VanessaE: by little ugly icon, yeah |
09:38 |
thexyz |
that's why I suggest to use gravatar |
09:38 |
thexyz |
it can generate pretty patterns |
09:38 |
VanessaE |
not supporting something as simple as an avatar, it makes a site look incomplete |
09:38 |
thexyz |
reddit is incomplete |
09:38 |
thexyz |
and hacker news |
09:38 |
thexyz |
shit |
09:38 |
VanessaE |
reddit sucks anyway :) |
09:39 |
kaeza |
+1 for gravatar |
09:39 |
VanessaE |
but reddit isn't a forum |
09:39 |
kaeza |
one avatar less to configure :P |
09:39 |
thexyz |
you said "site" |
09:39 |
VanessaE |
s/a site/the site/ |
09:43 |
sfan5 |
-1 for gravatar |
09:45 |
thexyz |
elaborate |
09:45 |
VanessaE |
wouldn't it be easier to just add a rule that says something to the effect of "Your avatar may not be substantially the same as anyone else's." |
09:45 |
thexyz |
no |
09:45 |
thexyz |
who'll you give minetest icon avatar to? |
09:46 |
thexyz |
and then, it doesn't matter as others will be angry |
09:46 |
VanessaE |
whoever had it first? (or to c55, idc who) |
09:46 |
thexyz |
alright, so who had it first? |
09:47 |
VanessaE |
actually the first person I noticed was pilz but that doesn't mean I'm right of course. |
09:48 |
VanessaE |
or give it to whoever originally drew it (erlehmann?) |
09:48 |
VanessaE |
(I forget) |
09:54 |
celeron55 |
i think just disallowing using the same avatar as someone else would be sufficient here |
09:55 |
celeron55 |
the point of an avatar is to allow quickly seeing who is who (or at least on forum.minetest.net it is) |
09:55 |
celeron55 |
so having the same avatar simply defeats the purpose |
09:56 |
celeron55 |
unless those having the avatar represent some meaningful group |
09:56 |
thexyz |
oh well, good luck with that |
09:56 |
celeron55 |
well this is ultimately up to whether the moderators care |
09:57 |
celeron55 |
i trust them to have the widest view on what works and what doesn't |
10:18 |
|
darkrose joined #minetest-dev |
10:26 |
|
proller joined #minetest-dev |
10:33 |
|
iqualfragile joined #minetest-dev |
10:54 |
|
Calinou joined #minetest-dev |
10:54 |
|
PilzAdam joined #minetest-dev |
11:32 |
|
Jordach joined #minetest-dev |
11:58 |
PilzAdam |
AFAIK I am the first one who had the Minetest logo as an avater |
12:03 |
thexyz |
see? it's happenin' |
12:03 |
PilzAdam |
what? |
12:04 |
thexyz |
nah, nothing |
12:04 |
thexyz |
I dislike the idea of taking away those avatars from some users |
12:11 |
|
serengeor joined #minetest-dev |
12:41 |
|
nore joined #minetest-dev |
12:47 |
|
psedlak joined #minetest-dev |
13:01 |
|
Akien_ joined #minetest-dev |
13:19 |
|
Weedy joined #minetest-dev |
13:19 |
|
Weedy joined #minetest-dev |
13:24 |
|
kaeza joined #minetest-dev |
13:41 |
thexyz |
we should make remote_media fetch files based on sha1 hash, not name; any objections? |
13:48 |
|
MarkTwain joined #minetest-dev |
13:49 |
|
MarkTwain left #minetest-dev |
13:59 |
thexyz |
now how to do that without breaking anything |
14:05 |
proller |
+1 and without re-downloading if file exists |
14:06 |
thexyz |
i thought it doesn't redownload |
14:09 |
thexyz |
i didn't post it here, but the plan is: make it fetch media by sha1 hash, make cdn.minetest.net a public static server, use free cloudflare plan to cache it |
14:09 |
thexyz |
then we can add it to default minetest.conf |
14:44 |
kahrl |
there should always be the option to download by name |
14:45 |
kahrl |
it allows server owners to easily set up their own remote media server |
14:45 |
thexyz |
well we can also code a script, similar to django's collectstatic command |
14:46 |
kahrl |
maybe |
14:47 |
kahrl |
though if the server owner sees a 404 for some sha1 and wants to fix it that's harder than seeing a 404 for a name |
14:47 |
thexyz |
he or she just runs collectstatic |
14:48 |
thexyz |
we can include name as a ?param |
14:48 |
thexyz |
I just don't see the reason to support both |
14:49 |
kahrl |
we should ask some remote media server owners what they think |
14:50 |
thexyz |
that depends on how're we going to present it |
14:50 |
thexyz |
> hey guys, here's a free fast media server for all your needs |
14:50 |
thexyz |
> hey guys, we're breaking this once again |
14:51 |
kahrl |
can everyone upload to cdn.minetest.net? |
14:51 |
thexyz |
nope, I'm not going to give access to everyone |
14:52 |
kahrl |
then some are going to want to keep their own server |
14:52 |
thexyz |
that's true, and they can |
14:52 |
thexyz |
and it'll be even easier |
14:53 |
proller |
make names by sha + symlink to real file names |
14:54 |
kahrl |
proller, that doesn't work |
14:54 |
thexyz |
i object |
14:54 |
thexyz |
what if original file is changed |
14:54 |
kahrl |
a real file names can have multiple different contents |
14:54 |
kahrl |
name* |
14:57 |
proller |
real file names for custom servers |
14:59 |
|
Calinou joined #minetest-dev |
15:03 |
|
hmmmm joined #minetest-dev |
15:45 |
celeron55 |
only hashes for sure |
15:57 |
|
troller joined #minetest-dev |
15:59 |
|
Calinou joined #minetest-dev |
16:12 |
|
NakedFury joined #minetest-dev |
16:53 |
|
iqualfragile_ joined #minetest-dev |
17:25 |
|
Miner_48er joined #minetest-dev |
17:26 |
|
neko259 joined #minetest-dev |
17:36 |
|
tango_ joined #minetest-dev |
18:09 |
|
iqualfragile joined #minetest-dev |
19:01 |
|
proller joined #minetest-dev |
19:28 |
|
EdB joined #minetest-dev |
20:16 |
VanessaE |
the media server failing to reliably send .x files (or the client failing to always receive them) needs fixed first |
20:18 |
VanessaE |
as for filenames vs. hashes, hashes are obviously better for making sure the client gets exactly the variant of a particular file that it needs, but maybe that's overkill? |
20:19 |
sfan5 |
hashes are hard to manage without some kind of tool |
20:19 |
kahrl |
has there really been any case where the server doesn't send the .x files? |
20:19 |
VanessaE |
on my remote media server, I solved that by copying over all the default stuff first, then the third party stuff over that, so the client always gets whatever a mod provides, if it overrides something in default. |
20:19 |
VanessaE |
kahrl: mine? |
20:19 |
VanessaE |
except I guess it's the client not using them |
20:19 |
sfan5 |
happend to me too |
20:19 |
sfan5 |
on VanessaE's server |
20:20 |
kahrl |
VanessaE: from what I saw in the logs, the client get active object messages before it has the meshes |
20:20 |
kahrl |
gets* |
20:20 |
VanessaE |
kahrl: right, corrected myself after remembering that. |
20:20 |
kahrl |
so it's really an independent issue from the remote media thing |
20:21 |
kahrl |
or, well |
20:21 |
VanessaE |
call it "less-dependent" ;) |
20:21 |
kahrl |
the client sends TOSERVER_RECEIVED_MEDIA when it starts fetching remote media; maybe it shouldn't do that? |
20:22 |
kahrl |
does anyone know why it was done that way? |
20:22 |
* sfan5 |
doesn't |
20:23 |
kahrl |
indeed, the server checks if the client sent TOSERVER_RECEIVED_MEDIA before it tries sending mapblocks and objects |
20:26 |
kahrl |
thexyz: any idea? |
20:29 |
thexyz |
kahrl: because `getClient(peer_id)->definitions_sent = true;` |
20:30 |
thexyz |
though I'm not sure why that happens before media is received |
20:31 |
kahrl |
thexyz: when that is set the server assumes the client has all media |
20:31 |
celeron55 |
kahrl: sending that at that point sounds like totally wrong |
20:31 |
kahrl |
celeron55: it does to me |
20:32 |
kahrl |
but when I first saw it, I thought, "that looks so wrong, there must a really good reason for it to be done that way" |
20:33 |
celeron55 |
i can't remember any reason |
20:33 |
celeron55 |
try it |
20:33 |
thexyz |
nah, i probably just fucked it up |
20:33 |
thexyz |
sorry |
20:34 |
thexyz |
should be somewhere around line 848 in client.cpp |
20:34 |
kahrl |
I'm rewriting that part anyway, so no problem |
20:34 |
kahrl |
(using the httpfetch module) |
20:35 |
thexyz |
we'll probably need separate domain for cdn |
20:35 |
VanessaE |
...which is fucking FAST |
20:35 |
VanessaE |
:) |
20:36 |
thexyz |
hm? |
20:37 |
VanessaE |
thexyz: his httpfetch code. Menche tested it with an empty cache - sucked down all the media off my remote server in like 30s |
20:41 |
thexyz |
oh that's good |
20:43 |
thexyz |
I guess "minetest-cdn.net" is good enough, right? |
20:44 |
VanessaE |
not sure if he had something of a rose tint to his glasses or if that's accurate, but it was fast for sure :) |
20:44 |
sfan5 |
why not cdn.minetest.net? |
20:44 |
thexyz |
sfan5: cloudflare requires to delegate domain to them |
20:45 |
sfan5 |
Tesseract: reconnect, it was my daily disconnect |
20:45 |
kahrl |
VanessaE: what was his media_fetch_threads? |
20:45 |
sfan5 |
woops, wrong channeö |
20:45 |
sfan5 |
channel* |
20:46 |
|
VanessaE joined #minetest-dev |
20:47 |
* VanessaE |
grumbles |
20:51 |
kahrl |
<kahrl> VanessaE: what was his media_fetch_threads? |
20:51 |
kahrl |
(Menche's) |
20:51 |
VanessaE |
8 I think |
20:52 |
kahrl |
ah, that's good then |
20:52 |
VanessaE |
either 8 or 16, I forget which setting he had in force at the time |
20:52 |
kahrl |
I don't really know if changing the default to 16 would be a good idea |
20:52 |
VanessaE |
leaving it at 8 is probably sufficient |
20:53 |
VanessaE |
users whose net+cpu allow for more can always turn it up |
20:54 |
kahrl |
it's also the question of whether they use any other net applications that need more than best-effort network |
20:54 |
thexyz |
kahrl: will you change it to use sha1 hashes? |
20:54 |
kahrl |
thexyz: probably |
20:54 |
VanessaE |
right |
20:54 |
kahrl |
16 TCP connection have a big initial/minimal congestion window |
20:54 |
kahrl |
+s |
20:55 |
thexyz |
kahrl: ok, also please add curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1); |
20:55 |
kahrl |
it doesn't follow that by default? curious |
20:56 |
thexyz |
I think it doesn't |
20:57 |
kahrl |
what should CURLOPT_MAXREDIRS be set as? |
20:58 |
kahrl |
maybe 8? |
20:58 |
kahrl |
I'll set these for all http requests then, no need to distinguish between them I guess |
20:58 |
VanessaE |
REDIRS as in http redirects? |
20:59 |
kahrl |
VanessaE, yup |
20:59 |
VanessaE |
is there any point in allowing more than maybe 2? |
20:59 |
sfan5 |
2 should be enough redirects |
20:59 |
celeron55 |
why not allow more? |
20:59 |
sfan5 |
otherwise it would take too much time |
20:59 |
thexyz |
I don't think there's much point in allowing more than one, but why not 8? |
21:00 |
celeron55 |
it's completely useless to not allow more if it's a matter of a single parameter |
21:00 |
kahrl |
I just don't want to set it to infinite, to avoid infinite loops |
21:03 |
kahrl |
sfan5: if the server owner uses a slow configuration that's really their own fault |
21:04 |
sfan5 |
hm, yeah |
21:09 |
|
VanessaE joined #minetest-dev |
21:11 |
thexyz |
support for multiple remote_media servers would be extremely useful |
21:20 |
|
MaxCZ joined #minetest-dev |
21:21 |
|
Jordach joined #minetest-dev |
22:03 |
|
kaeza joined #minetest-dev |
22:42 |
|
Weedy_lappy joined #minetest-dev |