Repository navigation
Fix: Error [ERR_DLOPEN_FAILED] - #1546
Conversation
9a42a44 to
ecb7eba
Compare
|
@hawkaii thank you for reporting. |
|
it works |
Merged from: anomalyco#1546 Author: @CaffeeLake Auto-merged by opencode-fork integration system
|
could you fix the conflicts? also can you describe what environment this is? |
ecb7eba to
eae1932
Compare
|
I am also seeing this in a Linux 6.10.14 aarch64 container running in Docker Desktop on an M1 Max MBP, in either the Docker VMM or Apple Virtualization framework. This did not occur when using an amd64 version of the same container, though the TUI slowed to an absolute crawl when running in that manner (with or without using Rosetta on Apple Virtualization as well). So it seems like just generally Linux aarch64 was the cause, but only after |
|
This may be related to the following Node.js behavior:
~/opencode % ls ./node_modules/tree-sitter-bash/prebuilds/linux-x64/
tree-sitter-bash.node
~/opencode % node ./node_modules/tree-sitter-bash/prebuilds/linux-x64/tree-sitter-bash.node
~/opencode %
~/opencode % ls ./node_modules/tree-sitter-bash/prebuilds/linux-arm64/
tree-sitter-bash.node
~/opencode % node ./node_modules/tree-sitter-bash/prebuilds/linux-arm64/tree-sitter-bash.node
node:internal/modules/cjs/loader:1846
return process.dlopen(module, path.toNamespacedPath(filename));
^
Error: ~/opencode/node_modules/tree-sitter-bash/prebuilds/linux-arm64/tree-sitter-bash.node: cannot open shared object file: No such file or directory
at Object..node (node:internal/modules/cjs/loader:1846:18)
at Module.load (node:internal/modules/cjs/loader:1423:32)
at Function._load (node:internal/modules/cjs/loader:1246:12)
at TracingChannel.traceSync (node:diagnostics_channel:322:14)
at wrapModuleLoad (node:internal/modules/cjs/loader:235:24)
at Function.executeUserEntryPoint [as runMain] (node:internal/modules/run_main:171:5)
at node:internal/main/run_main_module:36:49 {
code: 'ERR_DLOPEN_FAILED'
}
Node.js v22.18.0
~/opencode %I couldn't find a command to call |
|
I'm running Debian Bullseye on WSL2 and experiencing the same issue. It's been unusable for a few days now. I’d appreciate a fix—thank you! |
|
@lsj5031 you can rollback to v0.3.101 and set autoupdate false in the global config until it's fixed, thats what I did |
|
I've identified the exact cause.
~/opencode % ldd ./node_modules/tree-sitter-bash/prebuilds/linux-arm64/tree-sitter-bash.node
not a dynamic executable
~/opencode % ldd ./node_modules/tree-sitter-bash/prebuilds/linux-x64/tree-sitter-bash.node
not a dynamic executable
~/opencode %
~/opencode % ldd ./node_modules/tree-sitter-bash/prebuilds/linux-arm64/tree-sitter-bash.node
linux-vdso.so.1 (0x00007ffc25d73000)
libstdc++.so.6 => /lib64/libstdc++.so.6 (0x00007f8871bc8000)
libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f8871b9c000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f8871b98000)
libc.so.6 => /lib64/libc.so.6 (0x00007f88719a6000)
libm.so.6 => /lib64/libm.so.6 (0x00007f88718b8000)
/lib64/ld-linux-x86-64.so.2 (0x00007f8871f98000)
~/opencode % ldd ./node_modules/tree-sitter-bash/prebuilds/linux-x64/tree-sitter-bash.node
linux-vdso.so.1 (0x00007ffe2a794000)
libstdc++.so.6 => /lib64/libstdc++.so.6 (0x00007f5e05969000)
libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007f5e0593d000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00007f5e05939000)
libc.so.6 => /lib64/libc.so.6 (0x00007f5e05747000)
libm.so.6 => /lib64/libm.so.6 (0x00007f5e05659000)
/lib64/ld-linux-x86-64.so.2 (0x00007f5e05d39000)
~/opencode % |
|
I've been facing this same blocking issue on a raspberry pi 5 running bullseye as well. Happy to test the fix once it's ready! |
d355987 to
d26b139
Compare
|
This |
|
it's working but the i had to build it by running |
d26b139 to
b9e0178
Compare
|
@thdxr @adamdotdevin This PR is ready. 🙏 |
|
I can confirm that in my previously mentioned aarch64 container, without this patch on the
After applying this patch, bash commands work as expected again.
|
b9e0178 to
2ad722f
Compare
|
when is this fixed expected to be merged? |
|
@thdxr or @adamdotdevin - @CaffeeLake has graciously been keeping this PR up to date and it would fix a serious issue facing arm64 users. Unfortunately it doesn't seem that the node-tree-sitter dependency is nearly as active as you two so unlikely to be fixed forward over there anytime soon. 🙏 |
|
getting the same issue, but only with certain models. ie Qwen in LMStudio via proxy |
2ad722f to
cf370fe
Compare
|
@thdxr Thank you for checking. |
cf370fe to
6b5d6c7
Compare
|
I discovered that downgrading @thdxr Is this specification okay? |
cba46fb to
e5c86ab
Compare
|
Any progress on this one? It makes the eintire tool not usable for me:( |
@iliagerman |
|
@thdxr Could you please review it again? |
I've been using it on dev with this patch just fine to keep getting updates. So no, it's definitely still working here as well. |
e5c86ab to
f30cf89
Compare
|
I suspect i hit this today but the message is slightly different so here goes: Bash Build the demo app ┃ ResolveMessage: Cannot find module '/$bunfs/root/../../../tree-sitter-d8jpqavy.node' from '/$bunfs/root/opencode' NixOS on aarch64. |
To avoid confusion, please create any bug reports other than those that occur in this branch in Issues rather than this PR.
If I was to implement that feature in this PR, the change would likely cause the review to be rejected, further delaying the merge, so I won't address it in this PR. |
f30cf89 to
fbd0b78
Compare
|
PR looks good but did you check with an actual compiled version? |
|
just tested on linux-x64, looks good |
Reduce preview packaging traffic while typing pasted skills so the share editor does not repackage on every keystroke. Add an e2e check to keep the debounce behavior from regressing. Co-authored-by: src-opn <[email protected]>



Fix: Error [ERR_DLOPEN_FAILED]
Fixes: failing to run command on new update #1509
Fixes: bug: All commands return
No such file or directory#1670In arm64 environments,
tree-sitterandtree-sitter-bashmay not work properly, and the following error may occur when executing commands.This is binary issue on the library side Fix: Error [ERR_DLOPEN_FAILED] #1546 (comment), so
this patch temporarily avoids the error by skipping the use of the library and reverting to the previous behavior in environments wheretree-sitterdoes not workFix: Error [ERR_DLOPEN_FAILED] #1546 (comment)Fix: Error [ERR_DLOPEN_FAILED] #1546 (comment).Since opencode is completely unusable in environments where this problem occurs and the urgency is high, I am submitting a PR in this format, but if you know of a better solution I would appreciate your information.