Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • 15:46

    lazka on master

    openshadinglanguage: update to … Merge pull request #11671 from … (compare)

  • 15:46
    lazka closed #11671
  • 15:46
    lazka closed #11096
  • 15:45

    lazka on master

    assimp: update to 5.2.4 Merge pull request #11673 from … (compare)

  • 15:45
    lazka closed #11673
  • 15:44
    debashish-ghosh commented #3163
  • 14:08
    hgdagon edited #11673
  • 14:08
    hgdagon edited #11673
  • 14:07
    hgdagon opened #11673
  • 11:47

    github-actions[bot] on master

    Cygwin: add 3.3.6 release notes (compare)

  • 08:35
    al-cheb closed #2591
  • 08:19
    ivellioscolin commented #2591
  • 08:16

    dependabot[bot] on npm_and_yarn

    (compare)

  • 08:16
    dependabot[bot] closed #223
  • 08:16
    dependabot[bot] edited #223
  • 08:16

    dependabot[bot] on npm_and_yarn

    (compare)

  • 08:16
    dependabot[bot] closed #224
  • 08:16
    dependabot[bot] commented #223
  • 08:16
    dependabot[bot] edited #224
hvonreth
@hvonreth:kde.org
[m]
After going crazy I stumbled over msys2/msys2-runtime#77
is there an eta for a release with a fix
hvonreth
@hvonreth:kde.org
[m]
ah its part of the above branch
hvonreth
@hvonreth:kde.org
[m]
any chance I can get a preview runtime without building it myself? This issue appeared at an inconvenient point of time...
hvonreth
@hvonreth:kde.org
[m]
I'm now deploying the msys2.dll from the artifact
Johannes Schindelin
@dscho
Right, that's the way to get a custom msys-2.0.dll without having to build it locally.
Note that we're kind of waiting for Cygwin to resolve a couple of issues in preparation to releasing Cygwin v3.3.5, which we will subsequently consume.
hvonreth
@hvonreth:kde.org
[m]
thx
jarda-zaluznik
@jarda-zaluznik
<1480c1> Does anyone know if there happens to be a really weird convenient way to execute a linux binary from a windows binary running under wine (path conversions would be a nice bonus, but I can probably handle writing a program as a shim or something)? Something like a reverse msys-2.0.dll?

<1480c1> wait, nvm, somehow

#include <stdio.h>
#include <unistd.h>

int main() {
    printf("Hello, World!\n");
    execl("/usr/bin/uname", "uname", "-a", NULL);
}

this runs fine when I compile it with mingw-w64 and run it

<1480c1> Now I'm curious, if I copy /usr/bin/git to /usr/bin/git.exe, could cmake.exe autodetect it properly?
jarda-zaluznik
@jarda-zaluznik
<1480c1> doesn't seem like bash likes to be run under wine, bash: initialize_job_control: no job control in background: Bad file descriptor
jarda-zaluznik
@jarda-zaluznik
<1480c1> kwsysProcessesAdd function failed, seems cmake didn't like me trying to run Z:/usr/bin/git.exe
jarda-zaluznik
@jarda-zaluznik
<1480c1> unfortunately, even with my simple execvp wrapper, git.exe log doesn't want to show up
<1480c1> ah, it was because it was trying to exec less, --no-pager solved that

<1480c1> now I need to find out why it's printing like

[root@eeffbb15351b SVT-AV1]# git.exe -C /SVT-AV1 describe --tags --dirty --abbrev --always --match "v[0-9].[0-9]*" 
[root@eeffbb15351b SVT-AV1]# v1.0.0-14-g484d592f-dirty

using cat and od -c shows nothing weird

jarda-zaluznik
@jarda-zaluznik
<Biswa96> This may provide some ideas with wine https://github.com/mstorsjo/msvc-wine
<1480c1> my inspiration was that
<1480c1> but rather than msvc, it's msys2's stuff, minus bash and using archlinux's git since I can't get msys/git to work
jarda-zaluznik
@jarda-zaluznik
<Biswa96> Are you using archlinux?
<Biswa96> archlinux has mingw cross compiler.
jarda-zaluznik
@jarda-zaluznik
<1480c1> yeah, but I wanted to reproduce the same exact environment, gcc patch and everything the same, that I use in my Windows docker containers but on linux because Windows containers cause too much pain and their startup time and runtime is abysmal. After getting gcc setup, I was thinking of incorporating the stuff from msvc-wine so I could theoretically ditch the windows container all together but still have as close as an environment for it as possible.
<1480c1> Right now the main linux containers use ubuntu2004 and uses mingw-w64 gcc and cmake from ubuntu, so it would be hard to use msvc-wine since that requires patches on top of cmake to work properly, so instead I wanted to try to pull in a normal windows cmake, sort of
jarda-zaluznik
@jarda-zaluznik
<1480c1> Has anyone encountered something like running a program under wine and the program exiting, and then printing out stdout, if not piped?
https://cdn.discordapp.com/attachments/921754161907925052/972273311251832832/unknown.png
jarda-zaluznik
@jarda-zaluznik
<1480c1> seems git4windows' git.exe works if I install it through pacman, thanks @dscho
jarda-zaluznik
@jarda-zaluznik

<1480c1> so I realized an issue

root@f7dbaa2a8f98  /libvpx-src
$ git ls-files --stage configure 
100755 beea6503298d36245634eeecc5a9456e26ab9eaf 0       configure
root@f7dbaa2a8f98  /libvpx-src
$ ls -l configure
-rw-r--r-- 1 root root 25649 May  9 01:40 configure

the execute bit isn't preserved across the wine boundary, might need to do some weird grep xargs chmod or something

jarda-zaluznik
@jarda-zaluznik
<1480c1> seems the issue is when redirecting stderr to /dev/null inside a subshell
https://cdn.discordapp.com/attachments/921754161907925052/973042700859375636/unknown.png
<1480c1> git in this case is just /cmd/git.exe symlinked
Jinbo Wang
@testforstephen
Hi, I opened a Windows Git bash as my terminal, then ran a command like "/usr/bin/env key=value java myJavaApplication" there. When I checked the process tree of the current git bash terminal, I cannot find the env.exe and java.exe process there. Actually the env.exe belonged to a standalone process tree, and its parent process was a process that's no longer alive. Does anyone know why the env.exe process is not a child process of git bash terminal? Thanks.
lazka
@lazka:matrix.org
[m]
@dscho: "<corinna> I'll prepare a 3.3.5 later today"
@testforstephen: not sure.. I'd guess a side effect of how cygwin works. Why is that a problem for you?
Jinbo Wang
@testforstephen
@lazka:matrix.org: The backgroud is I use Git bash as VS Code default terminal. And In VS Code, I have an extension to trigger a Java command in Git bash terminal, then I need to get current running Java process pid programmatically and monitor its state. As a VS Code extension, we know the current git bash process pid, so we plan to use the git bash pid as parent pid and find Java process from its descendants. But we actually found that the java.exe and env.exe processes are not in the descendants of git bash. That's my use case.
if I run java.exe in git bash directly, it's a child process of git bash. But if i append a command "env.exe" before java.exe, then java.exe runs a child process of "env.exe", and "env.exe" is not a child process of git bash. I cannot use the process tree to find the target Java process.
lazka
@lazka:matrix.org
[m]
I'd guess it's a side effect of cygwin's fork emulation, and env.exe is a cygwin exe. I see the same thing in cygwin
you can ask cygwin devs
Jinbo Wang
@testforstephen
@lazka:matrix.org thank you for the suggestion. Do you know where is the channel to ask cygwin question?
Johannes Schindelin
@dscho

"<corinna> I'll prepare a 3.3.5 later today"

@lazka:matrix.org thank you!

@testforstephen https://cygwin.com/problems.html should help (They have an IRC channel, too, I believe)
lazka
@lazka:matrix.org
[m]
Jinbo Wang
@testforstephen
@dscho @lazka:matrix.org thank you!
Johannes Schindelin
@dscho
@lazka:matrix.org BTW cygwin-3_3_5-release is available.
I'll get a start on it, but might only finish tonight.
lazka
@lazka:matrix.org
[m]
ok, thanks
Johannes Schindelin
@dscho
lazka
@lazka:matrix.org
[m]
master now dropped 32bit support
Johannes Schindelin
@dscho
Biswapriyo Nath
@Biswa96
I thought OP is asking how to change that behaviour with MSYS variable :thought_balloon:
Johannes Schindelin
@dscho
Right, but since they're asking why git behaves in a way seemingly independent of the environment variable MSYS, the answer is that the symlink behavior is determined at git init time.