AD_Haiku_onPC has quit [Read error: Connection reset by peer]
AD_Haiku_onPC has joined #haiku
<divad27182>
Hello. Why do I need wifi firmware if I have no wifi?
jibsaramnim has quit [Ping timeout: 480 seconds]
jibsaramnim has joined #haiku
cassisian has joined #haiku
diver has quit [Quit: Leaving.]
xet7 has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
B2IA has quit [Quit: Vision[]: i've been blurred!]
B2IA has joined #haiku
B2IA has quit [Read error: Connection reset by peer]
B2IA has joined #haiku
<divad27182>
OK. I formerly tried an ext2 filesystem on a raw disk, and it worked. Now I am trying an ext2 filesystem in a partition, and Haiku won't see it. What partition types will it see?
jibsaramnim has quit [Ping timeout: 480 seconds]
<divad27182>
no... that isn't perhaps the right question. It sees it, it just won't mount it.
mmu_man has quit [Ping timeout: 480 seconds]
cassisian has joined #haiku
jibsaramnim has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
<scanty>
hmm.... i just had a thought. it would be nice if Tracker windows (when in single window mode) to have a 'homej' button.
<scanty>
'home'*
jibsaramnim has quit [Remote host closed the connection]
<divad27182>
Is there any way to verify what Haiku is reading as my partitions? I think it is reading the data wrong!
jibsaramnim has joined #haiku
<divad27182>
Apparently the problem lay in the partition size rounding. It wasn't, even to the kilobyte.
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
xet7 has quit [Remote host closed the connection]
OrangeBomb has joined #haiku
cassisian has joined #haiku
jibsaramnim has quit [Ping timeout: 480 seconds]
cassisian has quit [Ping timeout: 480 seconds]
bbjimmy has quit [Quit: Vision[]: i've been blurred!]
<pol>
hi, there used to be a menu in Tracker that would show a directory sub-structure if I drag and hold a file over it. Is that something that got disabled on hrev58859, because I discovered this feature on a version from January this year?
zardshard has joined #haiku
yann64 has quit [Quit: Vision[]: j'ai été flouté!]
cassisian has quit [Remote host closed the connection]
cassisian has joined #haiku
HaikuUser has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
HaikuUser has quit []
chilledfrogs has quit [Quit: connection reset by purr]
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
chilledfrogs has joined #haiku
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
erysdren has joined #haiku
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<divad27182>
pol: You can apparently do it in "Recent Folders"
cassisian has quit [Ping timeout: 480 seconds]
mmu_man has quit [Read error: No route to host]
<divad27182>
pol: for me, at least, it made a symlink instead of a move or copy... might be modifiers for that
HaikuUser has joined #haiku
HaikuUser has quit []
<thowe>
What is the state of IPv6 support on Haiku? I installed Haiku on my old machine at work, and it seems to think it should be auto-configuring, but it doesn't seem to recognize that is should configure anything on the local network. Local network has rt adv server on it with stateless flags, so it should SLAAC.
<divad27182>
thowe: I've been asking about that lately. Apparently no SLAAC, no DHCP6. also apparently only 1 IPv6 address
<divad27182>
and you may have to use manual configuration for IPv4 as well.
<divad27182>
OTOH, I have had it working at one point
<divad27182>
I know, disappointing.
<thowe>
So, it needs me to manually configure? IPv4 configured just fine via DHCP.
<divad27182>
apparently IPv6 autoconfiguration of either sort isn't there yet.
<thowe>
I know how to manually configure v6, but I imagine most people do not.
<divad27182>
agree...
<divad27182>
and apparently the auto/manual/disabled switch is common to both protocols. :-(
Anarchos has joined #haiku
<thowe>
I see. Well, I don't want to complain too much. The progress made in Haiku is impressive.
<thowe>
Still kicking myself for getting rid of my old BeBox.
<thowe>
Is the C++ library documented somewhere? I seem to recall it was part of the inspiration for Qt.
cassisian has joined #haiku
<divad27182>
unknown. personall, I find a lot of documentation lacking
<divad27182>
and apparently some of my IPv6 issues are with the host it is talking to. Drat. Why isn't it answer ping?
<divad27182>
thowe: Apparently one of the tricks: after setting the IPv6 configuration, REBOOT
cassisian has joined #haiku
<divad27182>
well... maybe not
<thowe>
OK. I made a $100 donation. That will have to do for now. My home office situation isn't very helpful at the moment.
<Anarchos>
thowe thanks !
<Anarchos>
thowe if every time i answer you, you donate 100$, please ask !!
<thowe>
Heh. I would rather be able to donate code, but in the past decade I've spent more time doing data munging code for work in scripting languages. My C/C++ is very rusty. I originally learned C++ to add code to a program called QBrew. A beer brewing recipe calculator. Learned some Qy4 in the process.
<thowe>
Qt4, rather
<Anarchos>
thowe i began slowly, just looking at code and some times tried to solves bugs which annoy me.
<thowe>
I used to own most of the Be programming books back circa 2001 when BeOS was my daily driver. Life got in the way.
<Anarchos>
i hadno books but i bought haiku and gobeproductive back then
<Anarchos>
i still have all the CDS and manuals (even the boot floppies)
<thowe>
Ah, yeah. Had Gobe. Also had a game called Corum III that I played a lot.
cassisian has quit [Ping timeout: 480 seconds]
ClaudioM has joined #haiku
<thowe>
I may actually still have a few bits of media in the shed. I'll look later since I have to pull out my big pot to scald a chicken.
<Anarchos>
Begasus[m] my usb keyboard doesn't work inside qemu. it sends garbage
AlaskanEmily has joined #haiku
coronavirus has quit [Ping timeout: 480 seconds]
neoncortex has joined #haiku
B2IA has quit [Quit: Vision[]: i've been blurred!]
B2IA has joined #haiku
<thowe>
There's clearly been plenty of work bringing IPv6 stuff into the B specific code... Maybe the ipv6 support shortfalls are just config and interface related(?)
<Anarchos>
thowe i never used ipv6 so can't answer
<thowe>
I haven't quite figured out how to use this code navigator...
<Anarchos>
thowe the one i put the link of ?
<thowe>
yeah.
<Anarchos>
thowe just click on links. I can help you in private if you need
<thowe>
Well, yeah. Like, trying to figure out where the netinet headers are
<thowe>
ah, but I see you can go to the dir. Guess that's not what I am looking for.
<Anarchos>
Begasus[m] i can't get the qemu version to use network. Is there a special option to qemu-system-x86_64 ?
<divad27182>
Another issue with IPv6: It seems to break pkgman
<Anarchos>
ipv4 is way more stable.
cassisian has quit [Ping timeout: 480 seconds]
HaikuUser has joined #haiku
HaikuUser has quit []
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
piku has joined #haiku
cassisian has joined #haiku
arjen_ has quit [Ping timeout: 480 seconds]
cassisian has quit [Ping timeout: 480 seconds]
<Anarchos>
i only get 'net unreachable' in qemu :(
B2IAb has joined #haiku
B2IA is now known as Guest21374
B2IAb is now known as B2IA
arjen_ has joined #haiku
Guest21374 has quit [Ping timeout: 480 seconds]
cassisian has joined #haiku
zard has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
xet7 has quit [Remote host closed the connection]
Babaj has joined #haiku
cassisian has joined #haiku
cassisian has quit [Ping timeout: 480 seconds]
lubo76buntu has quit [Remote host closed the connection]
cassisian has joined #haiku
nephele has joined #haiku
Anarchos has quit [Ping timeout: 480 seconds]
<nephele>
waddlesplash: on my 4k display, now with your patch... that mouse cursor will take some getting used to. it's so different xD
cassisian has quit [Ping timeout: 480 seconds]
<nephele>
divad27182: If you want a bug re-opened just comment to the effect (and why) It can be then reopened by someone else, you need some higher prifs for that iirc (which the devs have)
<divad27182>
turns out with all the reboots I have had to do, the closing reason may have been reasonable....
<divad27182>
I'm not sure the BSD group semantics are reasonable, but that is another story.
<divad27182>
Though comparing them, I'm not sure the SysV semantics are right either....
<nephele>
If it's in posix then we follow that
<divad27182>
I would prefer to see: follow the containing directory, provided you are in that group.
<divad27182>
posix in no reasonable excuse either.... and I suspect you don't in a lot of cases.
<nephele>
Not sure what you mean with excuse, if we don't follow posix then we fix it to align with posix when a report is opened
<divad27182>
as far as I am concerned, posix is a bug sanctifaction project.
<nephele>
Personally I think the whole Owner Group Other system is flawed, but that is another story entirely ;)
<divad27182>
well, you can do better... but ACLs get rather hairy...
<divad27182>
VMS had system/user/group/other and read/write/execute/delete
<nephele>
Posix is what it is, no matter how "cool" it is, we do follow it... In some cases we provide seperate apis you can use to acomplish the same thing with different or better semantics, but when we do not we follow posix
<nephele>
in this case you could use BFile or BEntry as apis to interact with this stuff instead, and that can give you other semantics, because those don't have to follow posix
<divad27182>
Personally, I think the biggest problem with posix, ISO C, ISO C++, etc....
<divad27182>
is that they are PRIVATE!
<divad27182>
terrible thing for a "Standard" to be.
<phschafft>
people seem to keep missunderstanding what POSIX is. POSIX is basically a try to collect what is out there and get people on a round table.
<phschafft>
it ISN'T a standard-first-implement-later thing as most other standards.
<phschafft>
it doesn't claim to be good, or the best way. it only tries to document what is actually implemented.
<Hanicef[m]>
yeah, it's a collection of specifications, but ogbs is the base that is considered the bare minimum
<nephele>
pretty much yeah... it still makes it easier for application devs if these different OSes atleast behave similar in some respects
<phschafft>
yes.
<phschafft>
and that is what POSIX is really good at.
<phschafft>
it is not meant to make OS better, it is meant to make it easer for application developers to know what to expect.
<divad27182>
well, they have enshrined bugs, when ANSI C tried to get it fixed...
<divad27182>
and it needs to be fixed.
<Hanicef[m]>
which bugs?
<divad27182>
string.h and memory.h
<divad27182>
string.h defines str*, memory.h defines mem*, neither should include the other.
<divad27182>
that was what ANSI C had to say about it.
<divad27182>
but there were a few symbols (like sizeof, but there wasn't a sizeof at that time) needed for both.
<divad27182>
and a desire not to have additional headers...
<Hanicef[m]>
string.h defines both mem* and str* functions in pretty much every implementation in existance
<Hanicef[m]>
it's even required by ogbs
<divad27182>
right... but it was a BUG in version 7.
<divad27182>
it was not intended.
<Hanicef[m]>
i fail to see how that makes it a bug, especially since it doesn't even define memory.h
<divad27182>
I think you are too young at this. :-)
divad27182 has quit []
Anarchos has joined #haiku
<nephele>
I don't see what age has to do with that...
<Hanicef[m]>
also, ogbs 8 was released in 2024, 1 year ago
<Hanicef[m]>
that is very recent in terms of specifications
dodo75 has joined #haiku
arjen_ has quit [Quit: Vision[]: i've been blurred!]
cassisian has joined #haiku
arjen_ has joined #haiku
<phschafft>
I would say there are more pressing problems whan which header defines what.
<nephele>
Like the terrible CLI interface >:(
<phschafft>
and when people come up with 'you are too young to understand' it normally means they are mad that the world has kept spinning (without them) ;)
zard has quit [Ping timeout: 480 seconds]
<phschafft>
like open() or like no hardlink support in bfs ;)
<Hanicef[m]>
what's wrong with open(), exactly?
dodo75 has quit [Quit: Vision[]: i've been blurred!]
dodo75 has joined #haiku
<nephele>
I wonder, was posix_spawn then actually a posix "invention" or did this also just standardize a mechanism existing in some OS?
<phschafft>
it is a try to get the DOS style API into POSIX.
<phschafft>
and I think it even tells you somewhere in the notes.
<nephele>
I'm confused why linux man page and wikipedia say this is only for embedded systems when it has some much better performance semantics in some cases....
<phschafft>
naturally a bit between the lines as standards normally do.
<Hanicef[m]>
erm, what? open() preceeds dos, so that doesn't even make any sense
<phschafft>
Hanicef[m]: I was answer to nephele.
<phschafft>
FILO ;)
<Hanicef[m]>
ah
<erysdren>
nerds
<phschafft>
Hanicef[m]: open() is a hell when it comes to race conditions. openat() is to be prefered. same for all functions that take a filename/path.
<Hanicef[m]>
:D
cassisian has quit [Ping timeout: 480 seconds]
<phschafft>
plus I personally think that filenames/paths as they are used today are generally a bad idea. openat() does a good step forward in that regard.
<phschafft>
you also get a good bit more performance with them. and there are a few things you cannot implement without openat().
<nephele>
file names are very good, for the user... :P
<phschafft>
nephele: like advanced bash syntax!
<phschafft>
best if they are combined!
<Hanicef[m]>
posix in general has a ton of thread-unsafe functions in it
<nephele>
shush, i'm trying to copy to the clipboard :)
<Hanicef[m]>
that's the consequences of using technology that preceds multithreading
<phschafft>
Hanicef[m]: I'm not talking about thread unsafe stuff.
<nephele>
it's too late for light mode websites phschafft
<Hanicef[m]>
yeah, but they're all prone to race conditions :)
<phschafft>
open() is broken for all systems in which a process might be interrupted, even if that interrupt is coming from the hardware.
<phschafft>
so basically for all practical machines.
pol has joined #haiku
pol has quit [Remote host closed the connection]
<phschafft>
nephele: strange. over here it is light mode but dimmed. but feel free to have a look tomorrow.
<nephele>
what do you mean with "but dimmed"?
cassisian has joined #haiku
pol has joined #haiku
* nephele
goes to test a 10 line commit with hopes of massively improved productivityx
<phschafft>
nephele: grey background.
Aedil has quit [Remote host closed the connection]
<nephele>
why can't i run history in bash and then click a line to get it in my buffer huh
<Anarchos>
nephele type "!" with the number of the history
<nephele>
that just runs the command instead .-.
<Anarchos>
nephele history | head -XXX
<nephele>
Man i hate shell scripts Anarchos, this is why. there is always some wierd hack to kinda get what you want. but no good UI direction anywhere
<nephele>
just copying the line and pasting it is easier
jmairboeck has quit [Quit: Konversation terminated!]
<nephele>
phschafft: uncomment some define for a trace, immidiently get build failures for printf formats... ouch :(
<nephele>
guess i'll have to fix that trace before i can get my trace
<nephele>
I've started my patch for the copying thing again. you should "soon" be able to clicp text labels in the OS then, to copy them easily... like mac adresses :)
<phschafft>
nephele: I have no idea why shells don't support mouse interaction. specifically in the current command (the one you are entering right now)
<pol>
It's when I try to copy in Tracker, it used to show the content of the destination folder, or be able to choose subfolder.
<pol>
Can I make a report of it missing on hrev58859 and still existed on hrev58522?
<nephele>
It's a documented feature, if it got removed it is a regression. It is here for me though?
<nephele>
Or is this only when you try and drop a file on a folder that it would open this menu?
<pol>
Yes, this is the scenario I'm thinking about.
<Anarchos>
nephele where did you get build failures for printf formats ? I must have made some on bootloader...
<nephele>
if it was there before (and was working as intended) then it missing now is proably a bug pol.
<nephele>
Anarchos: no, i uncommented a trace define in some header
<Anarchos>
nephele which one ?
<nephele>
it's not a problem you caused. our build bot also checks for that nowadays
<Anarchos>
ok
<nephele>
Anarchos: servers/app/EventDispatcher.cpp has a TRACE_EVENTS
<nephele>
> don't know how to make libsupc++.a
<nephele>
how can i fix this build error? what does it mean?
<Anarchos>
nephele maybe you forget -lsupc++ ?
<nephele>
I'm building haiku Anarchos
<nephele>
guess i'm gonna re-configure the build and do a complete clean one just in case...
Babaj has quit [Quit: Leaving]
cassisian has joined #haiku
<nephele>
I wish there was a better way for build systems to check if random symbols are available, it always takes so long for these tests to go through
<phschafft>
hm?
<phschafft>
I mean, if you know the target system and it's version, with all the strangenesses it may have, then there is.
<nephele>
phschafft: most build systems use tests that involve running the entire compiler chain
<nephele>
but that then, for each test tests if the compiler works, in addition to "is symbol X available"
<nephele>
surely there would be a better way to somehow ask the compiler it's internal state for each symbol it could encounter plus headers? instead of doing countless runs?
<nephele>
like, i understand that the current way just checks the exit code of the compiler... but there surely is some better way to do this?
<nephele>
is this just shell scripting hell?
cassisian has quit [Ping timeout: 480 seconds]
<phschafft>
nephele: yes, you could do something like that IF you know all the details on the system you're running on.
pol has quit [Read error: No route to host]
<nephele>
can't we do some compiler magic api to provide this data?
<phschafft>
if you want to know if some name is usable then you need to run the compiler. end of story. if you are sure it is actually a symbol and not something else then you need to know where it is, if you know that you can just look for that.
<nephele>
if the compiler provides it, use that
<nephele>
and if it doesn't fall back on the old system
<phschafft>
so you would again... run the compiler ;)
<nephele>
yes, but i would run it *once* give it the entire set of questions at one time, and get one set of outputs. maybe in json or some other "fancy" format
<phschafft>
which would mean that the compiler would need to be able to tell the results apart.
<phschafft>
I mean you can do pool tests.
<nephele>
Well yes, this would involve some kind of api of the compiler
<phschafft>
like test for 10 things in one go, and if they fail test them individually to see what failed and what didn't.
<phschafft>
nephele: my point is that it may be impossible for the compiler to tell your tests apart other than to do them as indivual units.
pol has joined #haiku
<phschafft>
and doing them as indivdual units is exactly what the current way is doing,.
<nephele>
my point is that "test the entire compile process" is *not* what these tests want to acomplish in the first place
<nephele>
You might want to test this once, sure. but for the rest of these tests it's just 90% noise or useless work
<Anarchos>
nephele oh you are working on unit tests ?
<phschafft>
I never said that.
<nephele>
I did not claim that you did phschafft
<phschafft>
my point is that those tests cannot be done without running everything.
<phschafft>
so it just doesn't matter if you want to test the rest or not, you need to do it to do what you want to do.
<nephele>
In the current iteration, yes. But i think that is stupid, and there should be an api where the compiler can simply answer the question without you having to produce a working a.out executable
<phschafft>
it's a bit like asking if a boat can float but without using water.
<nephele>
it's a bit like asking if your boat can float and testing it by throwing each invidual component in the water by itself
<phschafft>
no, that is what you try to do.
dodo75 has quit [Quit: Vision[]: i've been blurred!]
<nephele>
No
<phschafft>
e.g. if you don't include the right headers you would not catch any macro that might redefine stuff.
<nephele>
so?
<phschafft>
but if you load a header and that fails for some reason your unit can't be used to test anything else.
<phschafft>
so the compiler would need to throw away all of it's state.
<phschafft>
also the next test might result in a false posive with some unrelated header being loaded.
<nephele>
testing for symbols in a standard way should be totally possible
<phschafft>
so you need to throw away all the state for negative, and for positive results.
<nephele>
Sorry, phschafft, where did i say that the compiler should provide this api by inputting the previous tests as plaintext...?
<phschafft>
if you look for symbols. sure. but not if you look for 'looks like a function'.
<phschafft>
nephele: throwing all the state away is basically the same as restarting the compiler.
<phschafft>
that is my point. the only overhead is the startup of the compiler itself.
<phschafft>
and I suggested to do pool tests to improve that.
<nephele>
Not really. You have a huge overhead by a) doing this all the time on every build b) every project implementing this in a non-standard way c) no caching whatsoever
<phschafft>
they can be manually created to match the actual reasons for the tests, and then they can provide a lot more speed.
<phschafft>
autoconf implements caching? not sure the details, but I often see ... yes (cached)
<phschafft>
surely that could be improved. just saying that there is some caching.
<phschafft>
pool tests could be done, that would be a huge leap into the direction you want and it doesn't need updates on the systems nor the compiler.
Anarchos has quit [Quit: Vision[]: i've been blurred!]
<phschafft>
you could also patch things like autoconf to provide OS specific global caching. no problem with that.
<nephele>
I don't like shoveling abstractions onto a problem when it could be better solved with something that actually does what you want, instead of doing it as a side effect
<phschafft>
I'm sure you can hook into the current cache and preload it with a set of common basics.
<phschafft>
then I don't see what you actually want.
<nephele>
not make the tests even more complex?
<nephele>
build systems are already too complex
<phschafft>
so... you ask to do it the way it is currently done? which is the most simple way to find out if something works: try it!
<nephele>
that was a response to your statement about pooling tests anyhow, not about preloading autoconf caches (which could make sense)
<phschafft>
I mean it's puting the boat into water right now. I fail to see what test could be simpler than that.
<nephele>
It's finding out if a symbols is available as a side effect of compiling a programm that may or may not match that semantic. these tests sometimes fail for unrelated reasons, then get the result back as "no the symbol doesn't exist" and the build system has no way to catch this
<phschafft>
if simple is what you ask for. if you want simple and performant, then you might be out of luck.
<nephele>
this is neither simple nor performant
<phschafft>
you will loose portability, only two go, simple, fast, portable.
<phschafft>
and the point about build system checks is to make things portable.
<phschafft>
so simple or fast. pick one.
<nephele>
I don't mind super-cool-method only working on $distro that ships the package for it *shrug*
cassisian has joined #haiku
<phschafft>
so you want more complexity?
<phschafft>
you want to add another case to the tests. where the same tests run in different ways on different systems.
<nephele>
So, you want to put words in my mouth?
<phschafft>
no, I want to understand what you want.
<phschafft>
or maybe at this point I might just not care anymore.
<nephele>
I suppose you ment "wouldn't that lead to more complexity then?"
<phschafft>
I mean you ask to add something, so it will most likely not reduce complexity ;)
<nephele>
That is an opinion, sure. but that is something else than claiming i want something....
<nephele>
In any case, no, this would not really increase complexity. You can even implement the tests as a backend for the cool api if you want... it would just be a slow error prone aproach to implement it
cassisian has quit [Ping timeout: 480 seconds]
Skipp_OSX has joined #haiku
cassisian has joined #haiku
dalme has joined #haiku
nephele has quit [Quit: Vision[]: i've been blurred!]
<phschafft>
I very much fail to understand how writing a second way to process source code does not increase complexity immensely.
cassisian has quit [Ping timeout: 480 seconds]
nephele has joined #haiku
neoncortex has quit []
<nephele>
What is the correct way to see how long a "char*" is?
<nephele>
I have a fText of type char* and want to stuff it into the clipboard
<Skipp_OSX>
sizeof(fText) doesn't work? Should be 8 I assume, the length of a pointer on 64-bit.
<nephele>
phschafft: it would not process source code at all.
<Skipp_OSX>
or do you want the size of the string? that would be strlen(fText)
<phschafft>
nephele: then it does something other than the current tests.
<nephele>
yes exactly phschafft
<phschafft>
also what Skipp_OSX said.
<nephele>
Skipp_OSX: thanks
<nephele>
I am really confused why BStringView does not keep it's data in a BString but in a const char* :)
<Skipp_OSX>
A BString is a wrapper around a char*
<Skipp_OSX>
We do this in a lot of places, does it make sense to do so? Probably not.
<nephele>
but with utility functions such as .Length()... or the ability to pass it around directly without having to specify the size aswell :)
<Skipp_OSX>
Yeah well you're expected to re-wrap it on your end.
<Skipp_OSX>
We don't want to pass around the extra overhead... I guess
<nephele>
Meh.
<nephele>
Skipp_OSX: working on the patch to let StringView let you copy it's content
<Skipp_OSX>
More than likely Be engineers were crunched for time and we never changed it is the real reason but let's go with the overhead thing.
<nephele>
haha :)
<nephele>
But as i heard "but BeOS did this!"...
cassisian has joined #haiku
<nephele>
maybe this should go on the list of things R2 wants to change? though i don't see any reason why we can't switch this to BString by adding a second constructor, and a second method to return the BString (and keeping the previous like it is, but unwrapping it...)
nephele has quit [Quit: Vision[]: i've been blurred!]
<Skipp_OSX>
kicked for heresy
<Skipp_OSX>
(j/k)
nephele has joined #haiku
<nephele>
I rebooted to test my change
<nephele>
though i guess i could have just preloaded libbe.so... my bad *shrug
<nephele>
Skipp_OSX: should I implement drag and drop in the same commit?
<Skipp_OSX>
uh, probably not
<nephele>
I mean to the one where you can copy the text
<pol>
Can somebody answer that: "when I move a file over folder the drill-down menu appears", but, is it necessary to release the mouse button or simply moving mouse up or down and loosing selection on the menu is enough?
<Skipp_OSX>
I see, well I don't see, what are you dragging and dropping? The text of a string view?
<nephele>
Skipp_OSX is the resident tracker expert
<Skipp_OSX>
I guess so yeah
<nephele>
Skipp_OSX: yes, for example the MAC adress in preferences/network
<phschafft>
nephele: so you want two sets of tests, one for systems that support your new way, and one for systems that doesn't?
<Skipp_OSX>
I mean, you could make them uneditable but selectable text views if you want to copy them
<Skipp_OSX>
but I understand, you want a larger feature, yeah I guess same commit if you want to implement text snippets
<nephele>
No selections
rakka has quit [Remote host closed the connection]
<nephele>
You can just click them now to copy them
<nephele>
(that part is already done)
<nephele>
now wondering if i should do drag&drop aswell, or if that is too much for one commit
<Skipp_OSX>
uh... I didn't know you could drag and drop like that
<nephele>
haven't really done drag&drop yet so unsure of how the api looks...
rakka has joined #haiku
<nephele>
You click it and it copies the text to the clipboard
<Skipp_OSX>
maybe double click to copy?
<nephele>
why?
<Skipp_OSX>
You'll clobber the clipboard
<nephele>
double clicking is already a mistake where we have it now
cassisian has quit [Ping timeout: 480 seconds]
<Skipp_OSX>
right-click => copy?
<pol>
But, if I navigate the subfolders and move a mouse down from there the file is moved without me releasing the mouse button https://0x0.st/80Dx.png
<nephele>
I could do middle click or right click instead
<Skipp_OSX>
no middle click is right out
<nephele>
but i don't think it's neccesary
<Skipp_OSX>
I mean, I never use middle click for anything, double click would be better.
<nephele>
double click is bad
<Skipp_OSX>
we double click a lot
<nephele>
how so?
<Skipp_OSX>
I know there is gnome setting that gets rid of it, maybe that's a good idea.
<nephele>
i think only tracker uses double click
<pol>
I was thinking it is a bug but I started to think it is an intended behaviour.
<Skipp_OSX>
double click to bring up all windows of app
<nephele>
wtf?
<nephele>
how is that discoverable?
<Skipp_OSX>
it's not nobody knows about it and it's terrible but that's how it is
<nephele>
the user guide sais double click is a skill you can expect a user to have, but i think this should really be in the might have or better the don't use category....
<Skipp_OSX>
"Double-clicking an application's entry already serves to bring the application to the front. This works whether it has one window or many. Admittedly this behavior is undocumented, but it's there in the code and works just fine..."
<nephele>
I know many old people who have massive troubles with double click, and the workaround of increasing the time for it is not a good solution...
<nephele>
Skipp_OSX: is this refering to if you have the application expander turned off and click on the header?
<Skipp_OSX>
my fear is that you're going to accidentally clobber your clipboard by single clicking on a text having no idea that makes a copy
<Skipp_OSX>
yeah
<nephele>
the cursor changes to indicate that you will make a copy
<nephele>
Skipp_OSX: wtf? then that should be merged imidiently :D
<Skipp_OSX>
"The double-click-activate mechanism is one click too many." is what I said
<Skipp_OSX>
but I don't make the rules
<pol>
I use Deskbar in expanded mode so no need to duble click.
<Skipp_OSX>
I guess if the mouse pointer changes that's ok
<Skipp_OSX>
the user is given feedback that it'll do something at least
<Skipp_OSX>
I could also accept right-click context menu with just the one item "Copy"
<nephele>
Skipp_OSX: the copy mouse cursor graphics is pretty bad though
<nephele>
it seems only tracker uses this cursor in one instance
<pol>
When dragging a file to different folder with some key pressed?
<pol>
Can't find what key it is when you left click and drag to different folder - there used to be a cursor changing to + sign.
cassisian has joined #haiku
<nephele>
Skipp_OSX: seems BAlert uses a BTextView, so i guess i should either add this functionality there, or make a child class of BTextView for BAlert?
<Skipp_OSX>
BTextView makes it wrap
<nephele>
yeah
<Skipp_OSX>
If we had a wrapping StringView we could use it in BAlert and other places
<nephele>
want to build one? :)
<nephele>
I thing BStringView could probably be adjusted?
<Skipp_OSX>
there is one somewhere...
<Skipp_OSX>
but yeah that's the idea, allow BStringView that wraps and use it.
pol has quit [Remote host closed the connection]
<Skipp_OSX>
I guess you could make your feature work on BTextView as well if not selectable
<Skipp_OSX>
!IsSelectable()
<nephele>
I suppose that could work
<nephele>
But i don't know if this semantic makes sense per default for unselecatable BTextView
<nephele>
Skipp_OSX: finally, i can copy the damn hrev number without having to type it off xD
<nephele>
drag&drop would be even cooler, if i can safely drag&drop it into a web view... without webpositive deleting the page and loading whatever i dropped
<Skipp_OSX>
you could drag and drop to create a text snippet
<nephele>
I don't quite understand your sentence in context, you mean dragging from a label into the webview?
<Skipp_OSX>
like if you drag and drop some editable text to Desktop you'll see
<Skipp_OSX>
select text, drag it
<Skipp_OSX>
It creates "Untitled clipping"
<nephele>
yeah, doesn't work with webpositive
<Skipp_OSX>
sure it does
<nephele>
I just tried it
<Skipp_OSX>
try it from the URL bar
<nephele>
I ment dragging into the WebView
<Skipp_OSX>
not from inside the web view no, that's doesn't work
<nephele>
like beeing able to drag the haiku version into the webview into a textbar
<Skipp_OSX>
alright but now you see the clipping feature.
dalme has quit []
<nephele>
I saw it, yes
<nephele>
Skipp_OSX: can you give me an example of a text/plain drag& drop in the code?
<nephele>
then i can maybe add it to BStringView in this commit aswell
cassisian has quit [Ping timeout: 480 seconds]
<Skipp_OSX>
in BPoseView::CreateClippingFile
<Skipp_OSX>
called by BPoseView::HandleDropCommon in Tracker
ChaiTRex has quit [Remote host closed the connection]
ChaiTRex has joined #haiku
<nephele>
huh? but that only accepts a drop?
<nephele>
i'm trying to send a text/plain string
<Skipp_OSX>
if (dragMessage->FindString("be:types", index, &type) != B_OK)
<Skipp_OSX>
so ultimately it pull from be:types I guess
<nephele>
Doesn't help me.... still have no clue how to implement the drag and drop like this
<Skipp_OSX>
well you asked for the code I gave it to you
<nephele>
no, that's code how to receive a drop in tracker and save it to disk
<nephele>
that doesn't help at all for implementing the sending side... i thought this was negotiated aswell?
<nephele>
tracker for example also draws the thing you are dragging while it's dragging, i would also need to do this somehow....
<nephele>
it makes the system feel so much more responsive ^_^
<Skipp_OSX>
so src/servers/app/CursorData.cpp is where you'd add more vector cursors
<nephele>
oh btw
<nephele>
tracker should switch to cursor grab once you do mousedown on a selection
<nephele>
and then to grabbing once you move further
<nephele>
Skipp_OSX: the cursors were always vector also, in data/artwork/cursors
<Skipp_OSX>
hmm well they probably didn't need to be redone then but oh well
<nephele>
they weren't redone
<Skipp_OSX>
could have grabbed the data from the vector in data/artwork directly
<Skipp_OSX>
well not redone but copied
<nephele>
Well, yeah. the build system could have done some automation here
<nephele>
but then IOM -> HVIF
<Skipp_OSX>
you can create a resource from a vector icon directly
<nephele>
we don't have anything to do this automatically...
<Skipp_OSX>
BResource does this
<nephele>
should Menus also have a cursor?
<nephele>
to indicate you can use them?
<Skipp_OSX>
I mean, no
<Skipp_OSX>
most things don't have cursors just use the regular hand
<nephele>
yes. currently
<Skipp_OSX>
right so is that a problem?
<nephele>
Yes
<nephele>
I noticed this very heavily in webkit2. everything there was hand and the websites felt really wrong, alone because the link cursor was missing
<nephele>
and i noticed that adding more cursor variations to the system makes it much easier to read the UI in places
<nephele>
(because the system acknowledges you could do something)
vdamewood has joined #haiku
<Skipp_OSX>
I tried to add cursors to Deskbar for the window width and to move around but I couldn't get it to work the cursor would revert to normal out of view...
angrystar17043 has quit [Read error: Connection reset by peer]
angrystar17043 has joined #haiku
<nephele>
Why would it stay the cursor when leaving the view?
<Skipp_OSX>
stupid tumblr tries to prevent me from getting the video source directly but they can't stop me
<nephele>
MediaPlayer fucks up with that video... but QMplay2 plays it okay
<nephele>
Skipp_OSX: tracker has the oposite problem
<Skipp_OSX>
hmmm but either way cursors have problems out of view, I'm sure it's solvable
<nephele>
honestly, i think this calls for a new api, tell the app server to remporarily overwrite the system cursor, and then return it to normal when you release the lock
<Skipp_OSX>
maybe Tracker is overwriting my cursor that's why
<nephele>
trackers column resizing stuff has the <-> cursor only if you hover over the column resized
<nephele>
resizer
<Skipp_OSX>
yeah the cursor management is problemeatic
<nephele>
if you move down then it becomes whatever it thinks is apropriate there
<nephele>
i think it should be MouseDown-> tell app server "only <-> cursor until further notice" on MouseUp -> tell app server "release cursor"
<Skipp_OSX>
yeah because it's only setting the cursor in TitleView
<nephele>
that's fine. it should not manipulate the state of the other views
<Skipp_OSX>
well I mean if you're dragging a column and move your mouse into the pose view it should still show left/right cursor I'd think.
<nephele>
i think it's okay that it setts it in the titleview to indicate you can start a drag, but when you already started it it should have a way to overwrite the system cursor, and block other apps from doing it
<nephele>
yes, i agree
<nephele>
if i resize views in Vision it actually flickers when resizing the BSplitter
<Skipp_OSX>
I guess there is SetCursor on the BApplication instead of the BView you could use that might work
<nephele>
won't work
<nephele>
what if your cursor is above the window instead? :)
<nephele>
or another window is in-between?
<nephele>
I think this should really be a global cursor lock