How To Install The Gd Library On Ubuntu
Savage Doherty
I installed skype translting software CLOWN FISH in linux(ubuntu 14.04 LTS). But it was not working properly. In the clown fish website they have specified required library files for the proper working of the software. The following are the files. How do I install those files
When I build the project (using make), it works perfectly, and it creates the links correctly (I have checked it with the ldd command). However, when I try to install it (with make install), the generated file in the installation path is not linked against the specified shared library.
Using ldd again, I have checked that the shared library is not found by the library generated in the installation path, although it is found in the library generated in the build path. How could I fix this?
What happens is that CMake automatically sets the RPATH of your binary, to reference the shared library from your project build in $CMAKE_BINARY_DIR. (You want to test the binary against the library you just built, not the one you installed yesterday.)
Assuming that you are installing to a destination that is not in the system search path, your binary now no longer "sees" its shared library. Consult CMake RPATH handling to have CMake set the RPATH of your binaries to the installation path of your shared library.
I installed KiCad 6.0 on Ubuntu 20.04 as it was told here - Install on Ubuntu KiCad EDA by command line, when I want to add library as it was told here - Getting Started in KiCad 6.0 English Documentation KiCad
But i get an error telling me that it's impossible to find openssl-dev.Also after some googling, it seems that libssl-dev is equal to openssl-dev, is that true ? (apt-get found libssl-dev on my server)
If the likelihood that the dependencies for the version of a package that is in the release of Ubuntu (or other Debian derived arrangements) is the same as the deps for the version you are trying to build, you could run apt-get build-dep nginx or aptitude build-dep nginx - this will not install the nginx package but will instead install all those listed as dependencies (and their dependencies, as usual) which includes libssl-dev (the package that you are currently looking for).
In most cases this will allow the build of the other (presumably newer) version to be completed successfully, and it saves you installing each library and its header files one by one yourself. Even if there are new dependencies in the other version you are trying to build, build-dep is a good place to start as it means that you only have to manually install the extra new dependencies.
One thing to note is that if you are compiling your own copy because you want different build options rather than needing a different version for some reason, you may be better of compiling from the repository's source for the package rather than using the upstream sources directly. This SO question is the first useful page that came out of a quick search, though you are likely to find more detailed tutorials easily if you need that.
An other small thing to note: the packages installed as a result of apt-get build-dep will be marked as manually installed as if you have done this by hand as you are currently doing. That means you can't remove them all in one go (there is no apt-get unintall-dep or similar) - though that is no different from the situation you'll get from manual library/header installs anyway (I only mention the fact as some people expect there to be a one-step way to undo a build-dep operation, and there is not).
For use of CUDA (i.e. CUDA compiled application support, not development support), the only thing necessary to install on your machine is a proper GPU driver install. (This will also include nvidia-smi).
For applications written to conform to the runtime API (e.g. vectorAdd), the cudart library is required (and not installed by the GPU driver) but the default compilation option for apps compiled/linked with nvcc is that cudart is statically linked to the app. So in this case, nothing else is required other than the same GPU driver install which is sufficient for both driver API apps and runtime API apps where the cudart library is statically linked.
There is no official roadmap for you to do what you are asking. Of course it should be possible, with some thought given to what I have said above. The exact roadmap will depend on exactly what dynamically linked scenarios you want to support, and keep in mind that providing application support this way will only support applications compiled against the specific libraries and versions you provide. If you depend on the app developer to do the bundling, then of course they know which API version their app was compiled against, and can include the appropriate libraries. If you maintain your target machine with the latest GPU drivers, then this methodology will (should) support all applications, regardless of which CUDA version they were compiled against.
which removed not only the CUDA files but literally dozens of other packages CUDA depended on. This gave me a tiny Ubuntu, which I further reduced by removing stuff like Boost libraries, QT, etc. (A fresh Ubuntu install would have also been a good idea, but I already had a lot of setup in this instance).
Then at the end I downloaded the NVIDIA driver RUN package (not .deb) and used
Unfortunately, it turns out that even after removing GNUstep and the /etc/GNUstep/ directory, the install generated the same overwrite error. It looks as though the Eggplant installation creates the directory later attempts to overwrite the GNUstep.conf file. An additional error that appeared this time was:
Well, invisible directories aside, I decided to return to apt and then I remembered: Metacity; I forgot to log out and back in using Metacity desktop. I promptly did that and sure enough, the /etc/GNUstep/ directory was there. So, I promptly removed it and started again using Software Install:
Next, I searched for at least one of the libraries not found (Note: one of them turned out to be provided by libssl1.0.0 which is obsolete and no longer updated, something we need to return to later):
I should note in closing (although there is much to note for future updates) that in another VM I have of Ubuntu, after installation of Ubuntu 23.04 pretty much every library has been upgraded so Eggplant was not possible to install. This is fine since the target should be the LTS version, but even with the LTS Ubuntu there are some hurdles to jump over. However, once those were jumped over, all was good.
Hey Chris,
thanks for putting it together and pretty sure you had some fun while doing it as well as the contribution here.
I can feel the same which I had when getting EPF finally startet on Ubuntu 20.04 and 22.04
And good find with the LTS end date.
I like to install gource which requires libglm-dev installed on the system. I am using Ubuntu 10.10. Maverick. The newest sources of the headers only library I found is not available for apt-get. Afaik gource does not allow to configure the location of the library.
Where do I have to put the source files of the library on a Linux system? I could symlink the header file of the library to some system directory which is included in the path but that is not the way one does install libraries, right?
I have the longstanding problem that the install.packages() command will not install or update dependencies. This means I have to spend hours, sometimes six or seven hours, installing packages and all of their dependencies, and all of their dependencies' dependencies, and so on.
I'm wondering if the problem is that I have multiple libraries in .libPaths(). I read somewhere that this can prevent dependencies from being installed, because R doesn't know where you want to put them. Could this be the problem?
I'm most interested in specific code (in R or in the terminal window) that can get the job done and fix my problem, rather than general suggestions. My coding and unix shell skills are good but not as expert as many of you.
I have not seen this problem here before and unless someone has run across it, it will be hard to make specific suggestions without more information. Having multiple libraries on .libPaths() is normal, so I can't see how that would cause this problem.
I suggest you go to the CRAN list of packages (CRAN Packages By Name) and pick one you don't have and that has at least one dependency that you don't have. The abc package looked like a decent candidate, but you may see something better. Start a new R session and run
I'm no expert on this but it looks like some of your packages are up to date but were installed under an older version of R and must be reinstalled. What version of glue do you have? The latest is 1.6.2 and it was released 2022-02-24. If you have upgraded from R 3.x to R 4.x since then, you need to reinstall glue. I think R 4.0.0 came out in April 2020. Did you stay on R3.x for a while?
I think that's right, but the problem is it's not just one package. When I go to install glue, it has dependencies that also are outdated, and I get the same errors that those need to be reinstalled. In the end, I have to manually reinstall around 100 packages one-by-one. It's a horrible mess. Why can't R automatically reinstall those dependencies? Shouldn't dependencies = TRUE take care of it recurrsively?
I confess to being especially frustrated, because R does not experience these problems on Mac or PC. When I upgraded to R 4.0 on my Mac, I simply re-installed tidyverse, and R automatically reinstalled any dependencies that were out-of-date. These problems are only experienced in the Linux environment.
Alternatively, see the r2u package, which will give you binaries. Very quickly. ALL of them. With system library dependencies as well. All by the guy who has been lead on all things related to Debian on R for over 20 years. Read carefully the instructions at GitHub - eddelbuettel/r2u: CRAN as Ubuntu Binaries and skip the optional step 5. It should pick-up everything in your to_reinstall except for some oddball and stuff not on CRAN. Then you can compare the installed.packages() with the saved list.
7fc3f7cf58