freenode/#sbcl - IRC Chatlog
Search
15:05:07
phoe
Do you build SBCL for Windows in an unzippable package? I need to download it on a corporate environment where I am not allowed to execute MSI installers.
15:14:53
jdz
stassats: Compiling cl-freetype2 complains about invalid number of arguments to SB-KERNEL:THE*.
15:19:33
jdz
Similar errors involving SB-KERNEL:%SET-SAP-REF-32. It looks that in all cases SETF is also involved.
16:17:41
phoe
An exception occurred in context #.(SB-SYS:INT-SAP #X00227550): #.(SB-SYS:INT-SAP #X00228760). (Exception code: 1722)
16:20:04
phoe
https://stackoverflow.com/questions/6386856/process-crash-with-the-message-rpc-server-is-unavailable-on-opening-a-save-fil
16:24:44
phoe
"The only way that i found (not clean) to bypass this bug is to add the Win32 exception 0x6BA in the exception list of Visual Studio (CTRL+ALT+E) and disable it."
16:32:26
|3b|
phoe: is handle-win32-exception in the backtrace? if so, can you see the value of the CODE local variable in it?
16:34:24
|3b|
but that's where you would ignore it if you wanted to (though i don't think that should be ignored by default if it is an actual unhandled exception)
16:34:31
phoe
I'm trying to squeeze all I can from that stackoverflow answer while trying to understand Windows exception handling...
16:35:54
phoe
And it looks like it's possible to return from handle-win32-exception normally. https://github.com/sbcl/sbcl/blob/master/src/code/target-exception.lisp#L85 does so.
16:36:40
phoe
So, theoretically, I could add that exception code to this COND, cross my fingers, and tell SBCL to ignore it.
16:38:44
phoe
Because the description of https://sqlpowershell.blog/2016/05/25/the-rpc-server-is-unavailable-exception-from-hresult-0x800706ba/ is insane.
16:39:00
phoe
I want to open a file dialog and I get some weird Windows networking and remote computer shit.
16:40:19
phoe
stassats: signal a separate condition type that contains the exception code in one of its slots. I'll be able to handle exactly that condition type and invoke the restart in question.
16:43:13
phoe
I've been launching it from Qtools. Trying if going down to CommonQt changes anything...
16:49:42
phoe
It's 18:00 for me and I need to run from work in a few moments - I'll download, start the build, and test early morning tomorrow.
16:51:03
phoe
I've been reading that ticket and it seems the bug is in some shell extension that goes insane
16:53:13
phoe
This VM needs some more time before I can compile stuff on it. (I've been compiling exclusively on Linux for a while.)
16:54:49
phoe
Or (I feel terrible for asking for this) if someone here has a Windows machine with a toolchain nearby, I would ask them for building the most recent SBCL master and sending me the resulting binaries in a ZIP file.
17:01:11
|3b|
ACTION is building it with patch, --fancy, and 20gb heap for now, let me know if you want something different
17:25:12
|3b|
did you want/not want the dbghelp patch? (better backtraces for c functions, but not sure if there are any negatives)
18:23:15
scymtym
phoe: for linux, it archives the three most recent builds. e.g. https://ci.cor-lab.org/view/sbcl/job/sbcl-master/featureset=default,label=ubuntu_trusty_64bit/
18:24:06
|3b|
tests/run-compiler.sh doesn't have a case for Linux-ARM64, so doesn't pass -fPIC when compiling things
18:24:11
stassats
if i'm ever going to get around using Azure for testing, i'm sure microsoft can afford that
18:24:19
scymtym
phoe: you can try the "all in one zip file" at the bottom of https://ci.cor-lab.org/view/sbcl/job/sbcl-master-windows/label=Windows_7_64bit/ws/ *when is it not currently building*
18:24:52
|3b|
kill-non-lisp-thread.impure.lisp fails due to "relocation R_AARCH64_ADR_PREL_PG_HI21 against symbol `wait_a_bit' which may bind externally can not be used when making a shared object; recompile with -fPIC
18:27:23
|3b|
does seem like it should be a blacklist rather than whitelist, if some platform doesn't need it