-
Notifications
You must be signed in to change notification settings - Fork 3
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Build concept? #602
Comments
Some 3rd party dependencies are vendored as Git submodules, so you have to do |
Ok. Here is my incantation.
This fails as follows:
The full output of the command sequence is below.
|
This appears to be a glibc-devel issue? However I observe
|
For some reason, DeepState requires 32-bit version of libc dev files: glibc-devel.i686 |
If that doesn't work for some reason, the easiest would be to disable DeepState build in CMake files, fuzzing the library probably isn't your top priority right now |
Ok. Let me try that. The 32-bit version is not in the configured list of repositories.... |
I don't see the option to disable that in the README.md? |
Yeah, I was referring to "files" as in editing CMakeLists.txt, because there is no option for that yet |
So maybe https://github.com/laurynas-biveinis/unodb/blob/master/CMakeLists.txt#L370
change to
? Or maybe a few lines down
|
I can't figure out any way to really clean out the generated makefiles. Just fyi. The following does not seem to be enough. So I am just redoing things from the git clone.
|
This one |
I never tested in-source builds, and always do |
ack. i will try that. still hitting the deepstate dependency on that library with that edit to CMakeLists.txt. |
Updated incantation works around the lack of the glibc-devel 32-bit library. I will update again once I have this working.
Hits an issue with boost backtrace.
|
partial worked around for backtrace.h
Build progresses further, but fails on -lbacktrace.
|
I have spun up an Amazon Linux 2023 instance to come up with the recipe there |
I’ve been using AL2. But I could switch my environment.
…On Thu, Nov 21, 2024 at 23:28 Laurynas Biveinis ***@***.***> wrote:
I have spun up an Amazon Linux 2023 instance to come up with the recipe
there
—
Reply to this email directly, view it on GitHub
<#602 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AATW7YD5MUXERAG77Z4DUNT2B3MJFAVCNFSM6AAAAABSFUFWDWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIOJTGA2TCMRWGU>
.
You are receiving this because you authored the thread.Message ID:
***@***.***>
|
On AL2023 I got a successful build with:
then out-of-source build with -DSTANDALONE=ON -DCMAKE_BUILD_TYPE=Debug |
Release configuration works too. I'll see if I can test AL2 |
OK, so AL2 packages Boost 1.53, which does not contain Boost.Stacktrace and your logs above suggest custom-compiled Boost. I believe for me the best way forward is to make that dependency optional. |
For AL2, try the branch at #603, which will not try to use Boost.Stacktrace if it's not present (this of course will require removing your, presumably custom built, Boost) (Manually disabling DeepState is still required on that branch) |
For DeepState, I am testing #604 |
Build still fails with DeepState disabled on the #603 branch
My process was:
The cmake output is:
|
How does the CMake output look like? |
updated the comment above with the cmake output. |
This is not that branch - |
doh. ok. let me fix that :-). |
Sorry. Multi-tasking. Here is the complete output using a new checkout. Still fails on -lbacktrace
|
Something's not right, the CMake output does not have the added Boost.Stacktrace line. What does In any case I have also merged that branch to master, you might want try to pull and retry |
|
I have no explanation. The commit is correct, but the CMake output does not match that commit - there should be one of the two added lines in the output, but there aren't: Lines 304 to 308 in 3045f0a
How does the updated master branch look now? |
Pulled from master. rm -rf build/* Same outcome (build fail for -lbacktrace). I do see one of those messages in the output
Maybe we are looking for the wrong header file? Or maybe we should look for the library rather than the header file? |
Trying this hack
With this, the build proceeds. |
and the build completes!
So the issue is what we are "finding" to determine that the backtrace library is not available. By the way, here is what I did to install boost.
Not the exact commands, but how I did it. |
Ah, I see now. The check I added handles the case of installed AL2 Boost package not having Boost.Stacktrace but not custom-built Boost. If you don't need it for other things, just replace your build with Time permitting, I will see if I can work out correct link libs for AL2 with custom-built Boost |
Thank you VERY MUCH for your help with the build process! I will follow up on the thread-safe scan in December. |
I aggressively blew away my boost 1.8.5 install in /usr/local/{lib,header} and installed boost via yum as
I now have a build which works with only the single override to disable deepstate in CMakeLists.txt
|
make test passes
|
From a download of master, the following does not succeed.
cmake -DSTANDALONE=ON .
The detailed output is below.
Is the expectation that the user should check out each project into the corresponding 3rd_party library before invoking cmake at the top level? Or am I missing something and there is an incantation which will do all of this automagically?
Thanks,
Bryan
The text was updated successfully, but these errors were encountered: