SQLAPI++: Get path to shared library loaded by executable - ld

SQLAPI++ has an unusual feature where you set a string to tell it where to find the ODBC shared library. In my case this is libtdsodbc.so, and my application actually links that library at build time, but at runtime this is not enough for SQLAPI++ to work.
My code is:
SAConnection conn;
conn.setOption("ODBC.LIBS") = "libtdsodbc.so";
conn.Connect("SERVER=...", "", "", SA_ODBC_Client);
ODBC.LIBS is documented like this:
Forces SQLAPI++ Library to use specified ODBC manager library.
The above code works if you set LD_LIBRARY_PATH to a directory containing libtdsodbc.so. But if you don't, Connect() fails:
libtdsodbc.so: cannot open shared object file: No such file or directory
DBMS API Library 'libtdsodbc.so' loading fails
This library is a part of DBMS client installation, not SQLAPI++
Make sure DBMS client is installed and
this required library is available for dynamic loading
Linux/Unix:
1) The directories in the user's LD_LIBRARY_PATH environment variable
2) The list of libraries cached in /etc/ld.so.cache
3) /usr/lib, followed by /lib
It works again if you set ODBC.LIBS to a full path rather than just a filename. But how can the application know which path?
My application (outside of SQLAPI++) finds libtdsodbc.so via its RUNPATH which is set at build time. This path is not a system path like /usr/lib. I'd like to have SQLAPI++ use the same library which is loaded in the application at runtime.
One idea is for the application to inspect its own RUNPATH, search for libtdsobc.so, and use that path. But this requires quite a bit of fiddly code to basically reimplement what ld.so already does.
I don't want to bake the path into the executable at build time separately from RUNPATH, because I sometimes edit RUNPATH before deployment (and then I'd need to edit two things).
Ideally I would like to tell SQLAPI++ to just use the library which is already loaded. I can figure this path out by running lsof -p PID | grep libtdsodbc.so but running shell commands from within the executable is not a good solution (and again I would rather not reimplement lsof).

You could either use dl_iterate_phdr (the link also includes a sample code which prints out lib names) or manually parse /proc/self/maps.

Related

Deploying config files to PLC

Is it possible include arbitrary files (in this case a .csv) from a TwinCAT project direct to the Boot directory of a PLC?
By using PATH_BOOTPATH in the file open/read FBs it is possible to load files from this directory in a convenient manner regardless of whether using a CE or Windows deployment, However deployment of files to this location seems to be the sticking point.
I know that a copy of the project code is included within the CurrentConfig<Project>.tpzip file, but this file is not easily accessible from code, or updateable.
I've found the 'Additional Files' section within the system configuration, but it makes little sense.
Adding a file from inside the project as a 'Relative' path doesn't seem to do anything
Adding a file from inside the project as an external path includes the file (via symbolic links?) in the 'CurrentConfig.tszip' file, which has the same issues as the .tpzip
Adding an external file as an external path again includes the file inside of the .tszip.
I'm willing to accept that this might not be possible, but it just feels odd that the PATH_BOOTPRJ and PATH_BOOTPATH roots are there and not accessing useful paths.
Deployment
To quote Beckhoff:
Deployment is used to set up commands that are to be executed during the installation and startup of an application.
The event types are essentially at what stage of the deployment process the command is performed, where the command can either be copying a file or execution of a script/program.
Haven't performed extensive testing but between absolute/relative pathing and execution this should solve nearly all issues with deployment configuration.

In Adacore's GPR file, how can I set the compiler include search paths for C++

I have some C++ code that I need to compile using Adacore GNAT Programming Studio.
One file (SomeHeader.h) is in a Common directory (../../Common/) relative to my GPR file.
Our convention for C++ include directives is to use
#include "Common/SomeHeader.h"
No matter what I do, I cannot get GprBuild to find "Common/SomeHeader.h"
I followed the instructions here at AdaGem 108 with modifications for C++
for Include_Switches ("c++") use ("-I ../../");
and
for Include_Path ("c++") use "../..";
None of this seems to work for me during gprbuild and frustratingly I can't seem to get at the backend command that gprbuild is using even after turning the build verbosity up.
I see some temp files in the build messages but they get deleted before I can access them.
I am using Adacore GPS 17.1 on Windows 10 i686-pc-mingw32, GNAT Pro 17.1.
Does anyone know how to get include search paths working in Adacore's Gprbuild?
If you want to use relative paths, and you are dead set on using the -I flag, be aware that the current directory at the time you compile your c++ code it is set to the obj directory configured for grp.
So if the directory you want to include is located at C:\Foo\Bar\src\include\ and your grp obj directory is at C:\Foo\Bar\env\gpr\obj then your relative path will need to use -I..\..\..\src\include
I haven't tried to use gprbuild for compiling C++ source text yet, but I suppose it works more or less like with Ada, where you add the relevant directories to the Source_Dirs attribute:
project Cookie is
for Languages use ("C++");
for Source_Dirs use (".",
"../..");
[...]
end Cookie;

Relative path in Xcode generated DWARF file

When I compile my swift framework MyFramework I makes so that Xcode produces MyFramework.framework (the compiled framework) and MyFramework.framework.dSYM (the dSYM file).
By running dwarfdump myframework.framework.dSYM/Contents/Resources/DWARF/MyFramework I've noticed that all the paths to the source files are absolute paths.
This makes it impossible to share these two artefacts (together with the source code) with my colleagues and allow them to use the compiled framework and being able to debug their code properly. More precisely they are not able to step-in inside the framework with the debugger.
For this reason they need recompile the framework which is very inconvenient.
Is there a way to modify the DWARF to edit these paths or even better to convince Xcode to use relative path in the DWARF file?
While I'm not aware of a way to change the paths stored in the dSYM or to make them relative, I can offer a way to work around the issue and be able to debug with those symbols on an arbitrary machine with source code in a different base path.
Once you find out what the original path is (e.g. /Users/rob/src/Project), and you have the code for that library in e.g. /Users/luka/Sources/Project, then you can issue this lldb command, which will replace for that session of the debugger the original path with your path:
settings set target.source-map /Users/rob/src/Project /Users/luka/Sources/Project
Unfortunately you'll need to run this command for each debugging session. To do that, you can create a breakpoint in your main, which runs that command and continues execution without stopping.
To find the original source path you can either use dwarfdump as you mention, or you can look up a symbol you know is in that binary from lldb with im loo -v -n MySymbol or through other parameters of the command (depending on the type of symbol you're trying to use).

Is it Common to use MEF with a config file for specifying a plugin path?

I have heard that MEF reduces the need for creating config files, but if I have a few different plugin paths that vary depending on the client running the app, is it common and a good idea to have a config file that specifies the correct path. I want to avoid looping through all the DLLs.
Generally people have a well known plugin directory under the where the application is running from, i.e. \Extensions. However that said there isn't any particular reason you cannot do a configuration file for directories or exact extension assemblies.

How to distribute my Java program so that it is runnable by double-clicking a single file?

I have a Java rich client desktop app. that I want to distribute on some computers at work, but I've never done something like this before. People aren't too computer-savy at my workplace and since it is a student job, I won't be there for much longer and I'd like it if I could make my program easy to run by making it runnable when people double-click on it.
I also don't want to have to manually install a JRE to have it run. Basically, what I'd like to know is how to make my java application runnable easily by double-clicking (even if it's only on windows, it's okay). I'm pretty sure I'm going to need to package the correct JRE version alongside, but I don't know what's the correct way of doing this.
I read on some sites that you should not package a JRE along with your program because it makes people have multiple different versions, some of which are outdated, and it causes security issues, but this is not a problem in this case since the computers that are going to run my application are not connected to the internet and are only used to run this program anyway.
Somewhat related question: Since my application is currently an Eclipse project, I get my resources such as icons, images, SQLite database (for read and write), etc. using relative paths (e.g.: img/test.png).
Am I going to have to change any of those paths to have them keep working even while packaged?
What you're looking for is a JAR file. In eclipse, it's quite easy to make a Jar file. Specifically, you'll want to right click on your project, go to Export, and then select "Runnable Jar." Be careful with paths to folders. You may need to keep a resources folder next to the Jar file. You may need to provide some more specifics to get an exact answer on that. Typically, a Resources folder is located in the same spot as the JAR file (in the same folder on your computer).
A better option for easy install of a Java app. with a GUI is to launch it using Java Web Start. For the user, JWS is the 'one click' installation option that can (install & launch the app. then) add desktop shortcuts and menu items. A JWS launch would mean some more work for you, but it is a breeze for the end user.
To ensure a suitable JRE is present to run the app., use deployJava.js (see the JWS link for more details). The script would need to be reconfigured to get the JRE installer from your local network - the default is to get it from Oracle.
Most of the resources should be packaged in Jar files and supplied along with the app., but for the DB, use the JNLP ExtensionInstallerService to call the DB installer.
..Java Web Start is kind of a link (or I can make it a shortcut on the desktop) that the users will click to either install the JRE and run the program if the JRE isn't installed, or just run the program if the JRE is present on the computer.
The way it would work is to have a web page on the local intranet. When the user visits the page, the script checks for a suitable JRE.
If it is present, it writes the link to the launch file.
If there is no JRE, or the version is too low, it will guide the user through installing it (just a matter of them clicking 'OK' when prompted). Then it will put the link to the app.
I can then configure the link to grab the JRE from the server on our network.
That's the part where you need to reconfigure the script. AFAIR the script exposes an URL at which to look for JREs - that can be changed to point to a place on the intranet.
..So "Web" is only just in the name, the computers don't have to be connected to the internet to have this work, right?
Yes. JWS is a great launch technology for Java rich clients, but is a poorly chosen name.
To make the problem run by double clicking it you can distribute it as a jar file or a batch file to call the jar file.
For the installation part you can make a batch file that checks if java is present and then call the installer if it isn't.
Edit:
The batch code:
IF DEFINED JAVA GOTO ok
java-installer.exe
GOTO end
:ok
your-application.jar
:end
If you are finding it tough to implement the above mentioned methods. You can proceed with this simple approach.
Create a folder lib at a location. Place all the jars that your application uses into this. If you are able to create a jar for your application, you can very well place your application.jar into the lib folder too. Create a batch file at the same location that will contain the java command for your main class in it. The text within your batch might look something similiar to this :
set path="\lib\"
java -cp %path% package1.package2.MainClass
If you have any other dependencies, for ex: if you use images in your code under img/icon.jpg. Then you just have to shift the img folder to this location too.
Just zip these files using winrar and share it across. Running the batch file after extracting the zip would launch your java MainClass irrespective of the location in which it is placed in the client system.
PS : If you are unable to create a jar for your application and placing it in lib folder, just copy your bin folder with class files and paste it in the location and change the batch file accordingly to look for classes inside bin.