What happend:
I'm now working on a linux side-by-side installation issue since rpm no longer support to install the 32-bit and 64-bit same name packages on a machine at the same time. So I need to change the name of the 32-bit package from xxx.i386.rpm to xxx-32bit.i386.rpm. For supporting the upgrade of this package, I should provide the old package name: xxx in the spec file.
Problem:
I've looked into the pages of rpm.org looking for whether we can specify the version which we provided in the spec file like:
Provides: xxx = 16.0
I've done some test and it seems to work. But I can't find any offical explanation.
Does anyone else know the properly behaviour if I specified the provides package version in the Provides.
I'm putting aside that multilib is still supported by RPM (and I'm puzzled why are you doing that). But when i focus on that spec problem.
You have it nearly correct. Just omit that percent sign. Correct is:
Provides: xxx = 16.0
Note that the "16.0" should be just version. While it technically can be version-release, this will make you trouble some time later.
And if you want to provide really clean upgrade path, you should put there Obsolete too. See https://docs.fedoraproject.org/en-US/packaging-guidelines/#renaming-or-replacing-existing-packages
Related
I try to develop a thirdparty unixODBC driver, it is a secondary development based on the original file libodbc.so.2.0.0.
so I want to rename 'libodbc.so.2.0.0' to 'libodbc.so.2.0.0_renamed'. And soft link my dynamic library file to libodbc.so.2.0.0.
But I found an issue bothering me, when I rename native file and run 'sudo ldconfig', the file named 'libodbc.so.2' automatically linked to the renamed file 'libodbc.so.2.0.0_renamed', as below:
I could not understand that:
why it occurs;
how to appropriately replace the library.
I don't have enough ackownledge about linux, so that I failed to get any keyword to search and deal with it.
Could you help me, thank you very much!
Shared objects under GNU/Linux follow a specific version naming scheme, which is known by the loader (and OS component, actually part of libc framework) to determine if a newer library is retro-compatible with some older version to which a binary was originally linked against. By adding the renamed suffix, you are violating the convention and the dynamic linking system is getting confused. You should renamed as suggested by #Bodo above.
In addition, perhaps rather than using rename, you might consider using the very versioning scheme. From GNU Build System (aka Autotools) manual, the version cheme is like it follows:
Versioning: CURRENT:REVISION:AGE
CURRENT The latest interface implemented.
REVISION The implementation number of CURRENT (read: number of bugs fixed...)
AGE The number of interfaces implemented, minus one.
The library supports all interfaces between CURRENT − AGE and CURRENT.
If you have
not changed the interface (bug fixes) CURRENT : REVISION+1 : AGE
augmented the interface (new functions) CURRENT+1 : 0 : AGE+1
broken old interface (e.g. removed functions) CURRENT+1 : 0 : 0
Therefore a possible history of your lib might be:
1:0:0 start
1:1:0 bug fix
1:2:0 bug fix
2:0:1 new function
2:1:1 bug fix
2:2:1 bug fix
3:0:0 broke api
3:1:0 bug fix
4:1:1 bug fix
5:0:0 broke api
You might, for instance, call the older and newer versions of libodbc.so.x.y.z, according to your needs. Just an idea.
I need the source package for the mingw64/mingw-w64-x86_64-gcc-libs-8.2.0-3 package. I believe this is generated from the mingw-w64-x86_64-gcc source package. Looking in the repository I can find mingw-w64-gcc-7.3.0-2.src.tar.gz but nothing for gcc-8.*.
Most other packages have a simple relationship between binary and source package names. In a few cases (GCC is one) a single source package is used to generate multiple binary packages. However the naming is usually fairly obvious, and version numbers stay the same. I can't find any GCC-related source packages with the "8.2.0" version number.
Does anyone know where to get the source for the gcc-libs mingw package from?
You might want to look at here. You can find GCC 8.1.0.
Open an issue on https://github.com/Alexpux/MINGW-packages/issues
Apparently the maintainer forgot to upload the source package, or some script is failing.
I maintain a commercial, binary package for debian, ubuntu, & etc. and have gotten complaints about "bad package quality" from the Ubuntu installer. The root cause of this appears to be that the package installs files in the /opt and /etc/opt directories which apparently throws lintian into a hissy fit. As I understand the purpose of these directories, they are meant to serve the needs of "Add-on" applications although I have yet to see any definition of what "add-on" is supposed to mean. I have attempted to create a lintian override file for my package in the /usr/share/lintian/overrides directory and, when I do so, I get the following report from Lintian:
N: Some overrides were ignored, since the tags were marked "non-overridable".
N: The following tags were "non-overridable" and had at least one override
N: - dir-or-file-in-opt
Is there any way around this apparent obstinacy?
If you're using debhelper to build your package, lintian overrides for the binary packages should go in the debian folder with the filename [package].lintian-overrides. This ensures the overrides make it into the proper directory on package installation. (For source packages, overrides go in the file debian/source/lintian-overrides.)
That said, short of not putting your files in /opt, I'm not sure there's a way to completely get rid of the dir-or-file-in-opt lintian complaint. Though, I'm not sure it's necessary since, to my knowledge, the ubuntu/aptdaemon lintian profile (which disables the dir-or-file-in-opt tag) is used in most of the situations in which lintian would be run automatically during/before package installation. That's probably not quite the answer you were hoping for, but I hope it's helpful nonetheless!
Sources:
Debian New Maintainers' Guide, section 5.14
Lintian User's Manual, section 2.4
Cendio ThinLinc bug #5232
As far as I understand, "2" in "/lib/ld-linux.so.2" changes very rarely. There are old programs that can't work with new library and new programs that can't work with the old library. But they both refer to /lib/ld-linux.so.2, preventing installing both libraries.
BTW what are that version 'GLIBC_2.17' not found things? Why version is not in soname, but somewhere else?
Whenever a library is changed in a way that old programs do not work with the new version of the library this number is chaned. If old programs will still work with the newer library (and programs needing the new version will not cause a crash with the old library) there is no need to change the number because you can simply install the latest version of the library.
As far as I understand the "2" (it was "1" ten years ago) should not change any longer. The reason is simple:
The first version of "ld-linux.so.2" was written in a way that any version of "ld-linux.so.2" should work with any program using this file - maybe printing an error message when an older version of "ld-linux.so.2" is used.
This means: It is never necessary to install two different "ld-linux.so.2" files but it is enough to install the latest version of this file. You should not simply overwrite this file because it comes as a bundle with other files (e.g. "libc.so.6") that must have the same version.
In the "2" version of the loader .so files may contain a table of version numbers supported. So "libc.so.6", version GLIBC_2.16, may contain the information that "GLIBC_2.15" is also supported by this version of the library. (New versions should be backward compatible so this should be the case.)
Programs may contain a list of versions required (for example: a program requires libc.so.6 version "GLIBC_2.17"). The loader (ld-linux.so.2) checks if the library supports the version that is required by the program and refuses starting the program if not. In this case the error message form your question is printed by the loader.
The ~/.cabal/config stores configuration which cabal-install uses. I wanted to do some hackery on it. (Specifically, having multiple GHC versions installed, I wish to have separate documentation indexes).
I couldn't, however, find any documentation about its' syntax or variables except for what is included in default file. Is there any documentation available?
Edit: I've stated the goal above, but let me add some details: If the same package (eg. GTK) is installed in two versions of GHC they end up in the same documentation index file. I wan't that local documentation index to be separate for each GHC installation. I believe it is possible in some way using documentation directory setting, but there has to be a variable for currently used GHC version. If there isn't one there might be some workarounds available, but I won't be able to say that unless I see the documentation.
This seems to work, although I've only tested it with one version of GHC:
documentation: True
doc-index-file: $datadir/doc/$compiler/index.html
install-dirs user
docdir: $datadir/doc/$compiler/$pkgid
With the other options left at the default, this generates documentation in .cabal/share/doc/<ghc-version>/<package-name>, and the index in .cabal/share/doc/<ghc-version>/index.html.
There appears to be very little online - not even the haddocks for the cabal-install code. So your best bet may be to puzzle it out from the source. cabal unpack cabal-install, or view the cabal-install repo online. Look at SavedConfig in Distribution/Client/Config.hs. As an example, it imports GlobalFlags from Setup.hs; the individual flags, eg globalCacheDir, are associated with their config-file syntax (which is also the command-line syntax) in the globalCommand function below, remote-repo-cache in this case.
You should also ask dcoutts in the #haskell channel on irc.freenode.net, in case he has new docs available.