GMT broken when works with SQLite on Archlinux #393
Comments
|
This is a very mysterious issue on *nix. I have had a few reports of What appears to be happening is that the first |
|
Hmm, at this point you should be able to run a the |
|
The result shows below: I also check I have tried to solve this problem all this afternoon... In my situation, all data is stored in sqlite database, so I need to use SQLite to interact with them. I have changed all PyPlot code to GMT, so I also need to use GMT. And all these work fine on MacBook. Is there a way to fix the so files in GMT or compile the code when import (as it works when I only |
|
Actually it was not this output that I was expecting but rather the linux equivalent of Mac's |
|
How many of sqlite .so libs do you have? If you run the same |
|
The command I meant was |
|
All the packages were installed using It's too complex... I will try another Linux distribution today... Other related information: |
No, |
|
But what I wanted in one of my first posts is that you run the The idea is to try to see if it's using a different sqlite than that in /usr/lib |
|
I have tried gmt installed using I also tried Centos 8 (using docker), it seems the problem still exists.... the output of All the paths of libsqlite.3 in these outputs seem ok... but the problem exists... |
|
OK, tomorrow I'll install GMT and Julia in that Centos 8. But still, what about the?
|
in zsh:in juliaStill seems ok.... |
|
... running out of ideas. |
|
It seems SQLite.jl build its own so file in artifacts. but the version is the same as the one in However, when I ran nm at here no sqlite3_column_table. Is this the reason??? |
|
Almost for sure. But now the question is, why is that lib being used (indirectly via GMT) instead of the one in /usr/lib? I have this idea that when having GMT installed GDAL.jl, ArchGDAL.jl, netcdf.jl, HDF5.jl, GEOS.jl should all use the system libs that can be fished in GMT dependencies. This would save some GB of space ... and avoid problems like this. |
No, it downloads it. I think an issue should be open in BinaryBuilder or whatever because that lib is faulty. |
|
It's too complex.... 🤣 |
|
Have you tried to replace the lib in the artifact by the system one? |
|
It works... and then So it is definitely related to |
|
Yes, I don't know what will happen when |
|
After replacing the so file, all code works fine... Thank you!!! I think the so files are download from I am aware that most julia packages are maintained by volunteers, so sometime there are maybe some problems. (Maybe even julia itself, for example in the official coprs source https://copr.fedorainfracloud.org/coprs/nalimilan/julia/packages/, the build of latest version has been failed for months.....) I am closing this for the problems is related to other package.... Thank you again! |
|
Cool map. Are you using annotated lines for the parallels? |
|
The code I use is shown below. Generally, the figure is plotted using contour (solar zenith and geomagnetic latitude) grdimage (TEC data that I processed from GNSS observations), then some text, colorbar etc. Thanks for the simplicity of gmt and GMT.jl. It's very easy to produce this kind of high quality figures. it's nested in a long sequence of function calls, so it's difficult to exact the data... I'm sorry. |
|
Tell me, do you write these nice elaborated scripts only with the few man pages provided or you also look at the modules code? |
|
I read the document of GMT.jl extensively, and sometimes also gmt's document. At first, I will think of what elements I need then go to examples to find something similar and check the example code related to the figures. After plotting some figures (have some feelings of how GMT.jl works), recently, I just check the parameters of each function I need in (https://www.generic-mapping-tools.org/GMT.jl/latest/). Actually, I build the document locally (copy make.jl in doc then julia make.jl) so I don't need the internet to access the content. I think the most efficient way (at least for me) to learn this package is to run the examples in person, then change the parameters gradually to meet one's needs. I tried to use movie months ago but failed (🤣 )... If there is some update, I would like to have a try. Do you have some example code? |
|
I committed the movie.md in master. It's along manual and has two examples at the end that should +/- work but ... and at the end run, on the bash shell, the command that it prints. Vd=1 or no Vd at all should work too but I suspect it won't because the generated, |
|
OK, I will check the function, and try to use it in my code. I'm now using the following function to convert figures to movie after plotting all figures: By doing this, I need to think of all the detailed information, such as I need to place the title between colorbar and axis (so the differences of title wouldn't change the size of each figure.). And also it would be very slow if the figures are large... So I would like to try other ways. I will try to figure out how to use BTW, the |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I have found a very trick problem.
When I move some code to my desktop which runs Archlinux, GMT.jl will broke in some cases.
Changing the sequence of using
SQLiteandGMTcann't solve the problem. But if I runmakecptafterusing GMTthenusing SQLite, the problem will disappear.Another thing is all code works just fine on macOS (with brew).
At first I felt that the versions of
gdal(3.1.2 on macOS, 3.0.4 on Archlinux may be the key. But after compiling gdal 3.1.2 from sourceand replace the link in /usr/lib, the problem still exist.
As running
makecptafterusing GMTthenusing SQLitecan solve the problem, I think this may be related to some dynamic code generation inGMT.jlThe text was updated successfully, but these errors were encountered: