<Boosted>
It is for improve an existing app in HaikuArchives
<Boosted>
LockWorkstation, that without the filter, its purpose is easily defeated by calling Team Monitor
<Boosted>
and killing the app. It should make the user unable to call Team Monitor to kill the app, so only the authorized users with user-password can unlock the system
<Boosted>
This seems to be possible because the password-locking feature of screen savers can prevent to call Team Monitor while the screen saver is active and the lockdown feature is activated.
<phschafft>
hm.
<Boosted>
(I tried it).
<Boosted>
I tried to replicate it for LockWorkstation, but It does not fully work. It seems to block only one instance of Ctrl+Alt+Del, but no more than that, and team monitor then can be called (and the LW application then can be killed). Using a loop such as while could have unexpected results, like blocking any input from input devices and forcing the user to block the filter to re-gain the ability to input anything, to remove the bad f
erysdren has joined #haiku
<Boosted>
The code should work basically as this: while the target application is not running, it lets everything pass through it (B_DISPATCH_MESSAGE). But when the roster detects that the target application is run, then it should change to "active mode", filtering only the key combination that launches Team Monitor, other messages must be allowed to flow. If the target application is closed, then the filter must return to passive mode, le
<Boosted>
everything go on.
<phschafft>
hm.
<phschafft>
(please keep your lines a little shorter, two had their ends trimmed off)
<phschafft>
but now your problem is that it only works for the first message per kind?
<Boosted>
I think that's the case, but I'm not sure if it's once or a few instead of everything.
<Boosted>
I want to block a B_KEY_DOWN message. I guess...
<Boosted>
I'm not sure if I should also check for B_UNMAPPED_KEY_DOWN, B_MODIFIERS_CHANGED, B_KEY_UP or B_UNMAPPED_KEY_UP just in case.
<Boosted>
I also tried to detect if a "be:key_repeat" from B_KEY_DOWN is detected to block repeated requests of the same (multiple pressing) but without success on the result.
coolcoder613_ has joined #haiku
<phschafft>
maybe you want to add some logging for the path IsRunning() is returning false. maybe general logging for the case xxx:
coolcoder613 has quit [Ping timeout: 480 seconds]
<Boosted>
the logs added detected in the filter looper the mode change, and then
<Boosted>
"USER: lockworkstation_filter [Filter]: lockworkstation_filter: detected and filtered!"
<Boosted>
but after a second or so, Team Monitor could be summoned
mmu_man has quit [Ping timeout: 480 seconds]
Boosted has quit [Quit: Vision[]: i've been blurred!]
SLema_ has quit [Quit: Vision[]: i've been blurred!]
SLema has joined #haiku
SLema has quit [Remote host closed the connection]
SLema has joined #haiku
* OscarL
compares Python's errno.EPIPE value on Haiku (-2147459059) to the one on Linux (32)... "Surely this won't cause any issues at all, right?" he ponders.
<augiedoggie>
might need posix_error_mapper
* OscarL
writes that down.
<OscarL>
Thanks for the hint augiedoggie :-)
<OscarL>
On at least one test (test_broken_pipe) I'm getting: "AssertionError: 13 != -2147459059" (compares a returncode to errno.EPIPE).
<OscarL>
I'm *this* close to take a crapshot, and just negate the return value of that os.wait_status_to_exitcode() to see if it fixes anything, or totally breaks them :-D
<augiedoggie>
i'm not sure if the error mapper would take care of that one or not
<OscarL>
Welp, plain "os.popen("exit 42").close()" at least returns the same value on Haiku and on Linux: 10752 (== 42 << 8).
<OscarL>
so that os.waitstatus_to_exitcode() is doing something funky on Haiku :-)
<botifico>
[haikuports/haikuports] X547 262a3e2 - wayland_protocols: bump version
<OscarL>
down to only 9 open PRs... Begasus will put presure on me to either finish, or drop the 2 draft I have there :-D
ablyss has quit [Quit: Konversation terminated!]
OrngBomb has joined #haiku
OrangeBomb has quit [Read error: Connection reset by peer]
OrngBomb is now known as OrangeBomb
holdenbloodfeast has quit [Remote host closed the connection]
holdenbloodfeast has joined #haiku
<OscarL>
darn Python tests assuming hardlinks are supported :-(
Begasus has joined #haiku
<Begasus>
g'morning peeps
<OscarL>
hello Begasus.
<OscarL>
mmm my new Python 3.10.14 build didn't explode (even after my half-assed patching). Will it cause less tests to fail now? Only one way to find out!
<nekobot>
[haiku/haiku] 4b74781a000b - Catalogs: remove exr translations after purging from Pootle
<OscarL>
damn Pe using tabs for .py files :-(
Nephele has joined #haiku
x512[m] has joined #haiku
<x512[m]>
OscarL: It should be configurable.
<x512[m]>
Also Python with tabs works fine.
<OscarL>
x512[m]: yea, as long as you don't mix tabs and spaces in a way that makes it break :-)
<OscarL>
ideally, Pe should auto-detect if file is using tabs or spaces... but considering that Pe seems almost "abandonware" (and I suck at coding)... I should just remember to use the ExpandTabs extension :-D
<Nephele>
(or improve Koder ;)
<OscarL>
re: Koder... floating Find/Replace dialog? What is this... Pe? :-P
<x512[m]>
OscarL: Ideally Pe should use editor config file.
<x512[m]>
Guessing is always bad idea because it is unreliable.
<x512[m]>
Pe is better than Koder.
<x512[m]>
Pe is more feature-rich and 100% native software.
<Nephele>
x512[m], That's a stupid argument. Koder is also native software. "it's more feature rich" doesn't mean anything
<Nephele>
and it especially isn't an argument *against* improving Koder
<OscarL>
.editconfig would be nice, but I haven't had much troubles with simple editors doing auto-detect (and to be fair... .py is 99.9% of the time just spaces :-D)
<Nephele>
OscarL, what is that?
<Nephele>
OscarL, for the find panel there is an open ticket for Koder also
<x512[m]>
nephele: Koder is not native software because it use ported editor engine that is not originally designed for BeOS/Haiku.
<Nephele>
it's not native because it uses a ported library? By that definition not even app_server is native. That is ridicilous
<Nephele>
hell even the bootloader uses ported efi headers "not originally designed for BeOS/Haiku"
<x512[m]>
Pe can: compare files, show compile errors and jump to error positions, execute command, plugins and much more.
* OscarL
ways for Genio's new release... hoping it wont be too sluggish on slow PCs.
<x512[m]>
Koder is a simple wrapper over Scintilla.
<OscarL>
s/ways/waits/
<Nephele>
... So it's a IDE
<Nephele>
There is *no* reason to add command execution to Koder
<Nephele>
PE's bloated scope does not make it better in any way
<Nephele>
OscarL, why wait? wasn't "3.0" released recently?
<OscarL>
not seeing it on repos yet.
<Nephele>
... why the big announcement then?
<OscarL>
Pe without pipe extension would be lame. if that's "bloat"... so be it.
<Nephele>
I think it is kind of bad that they promoted it on mastadon as "The Haiku IDE" when it is *a* Haiku IDE... It gives the wrong impression that there is only one, or that it is somehow official
<Begasus>
g'morning nephele OscarL
<Begasus>
Hi x512[m] :)
<x512[m]>
Built Curl with HTTP 3 support.
<Nephele>
OscarL, The point is in Haiku you use severall windows for different things, Paladin is a great example. It does the project overview stuff. Command execution would be a second thing, and the text editor a third
<x512[m]>
Works faster than TCP.
JakeSays has joined #haiku
<Nephele>
that is an *actual* native IDE, not one that completely ignores haikus window managing semantics
<Nephele>
Hello Begasus :)
<OscarL>
nephele: re: multiple-windows: that might be something you like about Haiku. I don't.
<Nephele>
Why are you using Haiku then? stack and tile is one of the only big changes compared to other guis?
JakeSays1 has quit [Ping timeout: 480 seconds]
<Nephele>
do you only use fullscreen applications? In any case I stand by my point that an application designed for the OS semantics would be better
<OscarL>
Same as with any OS... some things I like, some things I don't.
<x512[m]>
Indeed.
<x512[m]>
Also different people have different opinions and demands.
<OscarL>
exactly.
<Nephele>
Exactly. It's very unfortunate to design these all-or-nothing apps then
<OscarL>
yahoo!!! "2 tests OK." LOL :-D
<Nephele>
Haikus gui is pluggable, if you ignore that you have a huge wasted effort, if there is just some parts i don't like in genio i can't just use something else, i can only fork the project and take over maintainership. Which is a terrible option
<OscarL>
now to double check results... I don't trust me.
<x512[m]>
About native status of app_server: all its interfaces and observable behavior are native, 3rd-party libraries are only used in internal implementation. Scientilla exposes public behavior, it is not the same.
<x512[m]>
Scientilla use Windows-inspired behavior.
<Nephele>
AGG can be used publicly too. Sctintall would be fine if you needleslly wrap it in a private api? xD
<Nephele>
Honestly your critzisim is kind of ridicilous. If it behaves in a non-native way then complain about that and it can be adressed but "it uses a library!!!" is not something that has anything to do with applications beeing native
<x512[m]>
Scientilla handles mouse clicks, keyboard, selection drawing. It is directly observed by user and Windows-centered. BeOS/Haiku native text editors react to mouse/keyboard in different way and also draw selection differently.
<x512[m]>
2D rendering is just math, calculation of pixel area coverage.
<x512[m]>
If AGG will be changed to Skia, it will look and behave the same.
<x512[m]>
Scientilla is the same as Blink to browsers.
<Nephele>
No
<Nephele>
Webkit draws the selection itself, and handles inputs. It is the perfect example of what you critizise. I don't see how that matters
<Nephele>
If anything behaves non-natively it can be fixed
<Nephele>
koder *definetely* uses native shortcuts for example. there is no ctrl-a or anything like that.
<x512[m]>
Webkit also have hard-coded behavior that can't be fixed without heavy patching.
<Nephele>
what behaviour are you talking about. I want concrete examples, not some vague idea of "this is shit"
<x512[m]>
I am not talking about Ctrl+a vs Alt+A. I am talking about more precise behavior.
<Nephele>
this style of conversation was already the problem on the forum recently, it's not nice here either. Properly back up your stuff with examples or leave it entirely
<x512[m]>
I never told in this discussion that non-native software is bad. I only claimed that Koder is not a native software.
<Nephele>
PE is not a native software because it uses bitmap icons
<Nephele>
See how that argument makes no sense?
<Nephele>
the truth is more this: both editors are in desperate need of maintenance
<Nephele>
And ideally we should all work on the same one ;)
<Nephele>
if PE can provide all the stuff that was the reason why Koder picked scintilla initially we can use it, but if it cannot we should improve Koder instead
* OscarL
at least sent some patches to Pe :-)
<OscarL>
still waiting on one PR... grrr :-P
<x512[m]>
Again, people have preferences, I prefer Pe, no matter is it native software or not. It have many important features to me. Koder is too primitive wrapper over Skintilla.
<x512[m]>
For example Pe have powerful search shortcuts where you can select search string without actually searching.
<x512[m]>
I sent some patches to Pe.
<Nephele>
Koders search is one example where it needs improvements indeed
<OscarL>
x512[m]: re: your Pe patches. Indeed, much appreciated!
<Nephele>
but that doesn't have much to do with preference, you can prefer the current state sure, but that doesn't say much about improvements in the future
<Nephele>
(since that would imply for example that since the shortcut preferences are bad now, we should *never* use them even if they are made better)
<x512[m]>
People who like Koder can send patches to Koder, people who like Pe can send patches to Pe. What is the problem?
<Nephele>
you build two basically equivalent softwares that will converge on a similar point
<Nephele>
it is just wasted effort
<Nephele>
Koder and Pe are not sufficiently different to be seperate projects and justify duplicating effort
<x512[m]>
It is totalitarian point of view that do not respect individuals opinion.
<x512[m]>
Koder and Pe are very different to me.
<Nephele>
that has little to do with opinion, they are designed deliberately to fill the same space
<Nephele>
Koder is designed to replace PE potentially
<Nephele>
Koder not supporting vital features has nothing to do with you "prefering" those features. it just means it is unsuitable at this time, but it does not mean it is not designed to replace pe
* OscarL
decides to go with a more conservative patch for Python 3.10.14, and see if that at least allows rebuilding of scipy 1.6.3.
<x512[m]>
If you are not know/use Pe features, it is your problem. It do not mean that features are not important to others and Pe must be abandoned according supreme leader the Nephele.
<Nephele>
What's wrong with you x512[m]? Do you really think it is apropriate to make "supreme leader" comments here?
<Nephele>
If you don't want to have a proper discussion then do me a favor and leave
<OscarL>
nephele: your tone is not particuarly pleasant sometimes.
<x512[m]>
You again make commands to me what to do like supreme leader.
<OscarL>
don't act surprised if you rub people the wrong way.
<erysdren>
nephele: dude, come on
<Nephele>
OscarL, That's fine. If you choose not to tell me I will not improve in those cases. That doesn't excuse behaviour of other people
<Nephele>
x512[m], I don't see how, if you refuse to engage in a discussion that is a "you" problem. It seems more that you are offended that I have a different opinion to you
freddietilley has joined #haiku
<OscarL>
chill out a bit.
<x512[m]>
"then do me a favor and leave" <- this
<Nephele>
I literally said above "the truth is more this: both editors are in desperate need of maintenance; And ideally we should all work on the same one ;)"
<Nephele>
And you misquote this as "Pe must be abandonded according to supreme leader"
<Nephele>
I never said Pe has to be abandoned, and yes, i do take offense to be deliberately misquoted
<x512[m]>
It is not ideally. Your ideal is like Ein Volk, Ein Reich, Ein Führer!
holdenbloodfeast has quit [Quit: Vision[]: i've been blurred!]
<x512[m]>
It is wrong claim that developers will work on what you want. Actually developer will work on what he want or do nothing. Unless you pay for a contract to him.
<Nephele>
I did not order you to do anything, i was simply having a discussion, or trying anyway.
<x512[m]>
"And ideally we should all work on the same one ;)", you prefer Koder, so Pe should be ideally abandoned. Simple logic.
* coolcoder613_
turns on the AC
<Begasus>
no way, Pe is still my main editor here, Koder is nice, but untill further notice, not any plans to use that over Pe
<Nephele>
No, your simple logic is wrong.
<Nephele>
I never said PE should be abandoned
* Begasus
hands coolcoder613_ some /DC to go with that ...
<x512[m]>
It is direct implication.
<Nephele>
No, it is not
* coolcoder613_
explains air conditioning
<Nephele>
and your behaviour based on your strawman is inexcusable
* OscarL
finds it weird coolcoder613_ didn't catch the AC/DC joke.
* coolcoder613_
points out the heat in #haiku with the AC remote
<x512[m]>
If "all work on the same one", what will happen to the other one?
* coolcoder613_
got the joke
<OscarL>
good, would be weird otherwise, being aussie and all :-P
* coolcoder613_
wasn't sure the acronym was universal
* Begasus
didn't get that AC thing :)
* coolcoder613_
was right
<x512[m]>
About Pe vector icons: toolbar HVIF icons are actually already exist, just not added to Pe source code yet.
<Nephele>
I don't care
<Begasus[m]>
/launches default apps ...
<Begasus>
lol, that doesn't work the same as in IRC :P
JakeSays1 has joined #haiku
JakeSays has quit [Ping timeout: 480 seconds]
<Begasus>
OscarL, managed the python tests?
<OscarL>
done some clean up there, at leat it seems I can consistently run the ones that give errors, without hangs or other weirdness.
<Begasus>
progress then :)
<OscarL>
Will know in 20 minutes again, after my next really-clean build (hp -c)
<OscarL>
mmm, now unpacking / git init + patch applying took 2.5 minutes only (half the time than earlier today/yesterday).
* OscarL
hope he didn't jinxed the build with that log filename: "hp python3.10 |& tee py3.10.4-build-final.txt"
<Begasus>
you'll find out soon enough :)
<OscarL>
unless PC goes kaput :-D
<OscarL>
fixed a test case, and hopefully, also the cause of that error we saw with numpy when trying to rebuild scipy.
vdamewood has joined #haiku
<Begasus>
oh! would be nice :)
<OscarL>
I'll give that one a try after one last run of "hp --test python3.10" (if I don't freeze first :-D).
<OscarL>
"test_kqueue skipped -- test works only on BSD" << Heh, will need to patch that to "or Haiku" after beta5
<Begasus>
module "org.kde.i18n.localeData" is not installed
<botifico>
[haikuports/haikuports] Begasus a55e41d - ki18n, revbump, qml file should end up in $dataDir/Qt6 (not in $libDir/qml). (#10695)
<Begasus>
you pushed some branch for python fixes?
<OscarL>
anyeway... Python 3.10.14 seems like an improvement already at least. /me opens a PR.
<Begasus>
okido
<OscarL>
done.
<Begasus>
no x86 patchset?
<OscarL>
mmm, pkgman search -D genio shows 3.0.0 available... but pkgman update genio gives me "Nothing to do" :-/
<Begasus>
resolving build dependencies failed: build-requires "devel:libeditorconfig" of package "genio_x86-3.0.0" could not be resolved
<Begasus>
whoops
<OscarL>
Begasus: re: x86 patcjset... no need anymore. "streamlined" it into the relevant .py
<OscarL>
(same as already done for 3.11/3.12)
<Begasus>
upgrade package genio-2.1.0-3 to 3.0.0-1 from repository HaikuPorts
erysdren has joined #haiku
<Begasus>
works here
<Begasus>
ah, still got one for the prior version, hence the confusien here
<Begasus>
confusion*
<Begasus>
OscarL, haven't tinkedered about tkinter anymore? (no pun intended) :P
<OscarL>
yeah, it is pretty dead currently. better keep it disabled, so possible users dont get deadlocked.
<Begasus>
k, just saw the message passing by during the build, that's why I asked :)
<Begasus>
Haiku's build hasn't changed python version right?
coolcoder613_ has quit [Read error: No route to host]
<OscarL>
not yet.
<Begasus>
k, need to be sure we end up like updating ICU74 :)
<OscarL>
even so... for building haiku they have a "build repository" with the needed packages (they don't auto-update to newer versions, only manually).
coolcoder613 has joined #haiku
<OscarL>
found why pkgman update was failing... conflict with llvm12_clang.
<Begasus>
well, some are mentioned in the haikuports repository file, I think those link to haikuports though?
<OscarL>
there's even a ticket that pkgman should print the error, same as pkgman install does.
<OscarL>
if I'm not mistaken... there's a separate repository that contains the build packages (separate from depot)
<OscarL>
thus why it needs manual intervention to update... need to change the files in Haiku's "repository" files, but also "upload" the packages somewhere.
<OscarL>
no clue about that pip+numpy error, sorry :-(
<Begasus>
bummer
<OscarL>
guess I'll try adding gast+beniget+pythran recipes, and try to build latest scipy with the numpy we have on the repos.
<OscarL>
darn scipy hates my guts :-(
erysdren has quit [Quit: Konversation terminated!]
<Begasus>
no matter what, we always seem to end up in python :P
<OscarL>
pkgman offers to install clang16, but we have clang18 available. wonder if the different names are preventing it to offer the newer one instead.
<Begasus>
nah, that was the issue last time
<Begasus>
it can use higher then llvm16
<OscarL>
yeah, but pkgman doesn't offers it, that's what I'm saying.
<OscarL>
(doesn't offers to insatll the newest that satisfies the >= 16 condition, I mean)
<Begasus>
nope, as llvm16 is the first candidate?
<Begasus>
got clang18 installed, had no problem updating genio
<OscarL>
pip wants you to install the lattest, pkgman.... meh...best I can do is the first one I find :-P
<Begasus>
heh
<Begasus>
why doesn't pip have a "search" option anymore? :/
<OscarL>
because some bots REALLY, REALLY abused that function, causing basically a permanent DoS attack.
<botifico>
[haikuports/haikuports] Begasus d9cb26b - kio, disable kcmutils (no mention found in the source, breaks kcmutils) (#10701)
<Begasus>
kcmutils dep for kio, kdeclarative dep for kcmutils and requires kio ...
<Begasus>
heh: Having this outside also avoids a circular dependency for now, because in a KCMUtils MR I want to use KIO::CommandLauncherJob, which will later be moved to KService