What is "vendoring" exactly? How would you define this term?
Does it mean the same thing in different programming languages? Conceptually speaking, not looking at the exact implementation.
Based on this answer
Defined here for Go as:
Vendoring is the act of making your own copy of the 3rd party packages
your project is using. Those copies are traditionally placed inside
each project and then saved in the project repository.
The context of this answer is in the Go language, but the concept still applies.
If your app depends on certain third-party code to be available you could declare a dependency and let your build system install the dependency for you.
If however the source of the third-party code is not very stable you could "vendor" that code. You take the third-party code and add it to your application in a more or less isolated way. If you take this isolation seriously you should "release" this code internally to your organization/working environment.
Another reason for vendoring is if you want to use certain third-party code but you want to change it a little bit (a fork in other words). You can copy the code, change it, release it internally and then let your build system install this piece of code.
Vendoring means putting a dependency into you project folder (vs. depending on it globally) AND committing it to the repo.
For example, running cp /usr/local/bin/node ~/yourproject/vendor/node & committing it to the repo would "vendor" the Node.js binary – all devs on the project would use this exact version. This is not commonly done for node itself but e.g. Yarn 2 ("Berry") is used like this (and only like this; they don't even install the binary globally).
The committing act is important. As an example, node_modules are already installed in your project but only committing them makes them "vendored". Almost nobody does that for node_modules but e.g. PnP + Zero Installs of Yarn 2 are actually built around vendoring – you commit .yarn/cache with many ZIP files into the repo.
"Vendoring" inherently brings tradeoffs between repo size (longer clone times, more data transferred, local storage requirements etc.) and reliability / reproducibility of installs.
Summarizing other, (too?) long answers:
Vendoring is hard-coding the often forked version of a dependency.
This typically involves static linking or some other copy but it doesn't have to.
Right or wrong, the term "hard-coding" has an old and bad reputation. So you won't find it near projects openly vendoring, however I can't think of a more accurate term.
As far as I know the term comes from Ruby on Rails.
It describes a convention to keep a snapshot of the full set of dependencies in source control, in directories that contain package name and version number.
The earliest occurrence of vendor as a verb I found is the vendor everything post on err the blog (2007, a bit before the author co-founded GitHub). That post explains the motivation and how to add dependencies. As far as I understand the code and commands, there was no special tool support for calling the directory vendor at that time (patches and code snippets were floating around).
The err blog post links to earlier ones with the same convention, like this fairly minimal way to add vendor subdirectories to the Rails import path (2006).
Earlier articles referenced from the err blog, like this one (2005), seemed to use the lib directory, which didn't make the distinction between own code and untouched snapshots of dependencies.
The goal of vendoring is more reproducibility, better deployment, the kind of things people currently use containers for; as well as better transparency through source control.
Other languages seem to have picked up the concept as is; one related concept is lockfiles, which define the same set of dependencies in a more compact form, involving hashes and remote package repositories. Lockfiles can be used to recreate the vendor directory and detect any alterations. The lockfile concept may have come from the Ruby gems community, but don't quote me on that.
The solution we’ve come up with is to throw every Ruby dependency in vendor. Everything. Savvy? Everyone is always on the same page: we don’t have to worry about who has what version of which gem. (we know) We don’t have to worry about getting everyone to update a gem. (we just do it once) We don’t have to worry about breaking the build with our libraries. […]
The goal here is simple: always get everyone, especially your production environment, on the same page. You don’t want to guess at which gems everyone does and does not have. Right.
There’s another point lurking subtlety in the background: once all your gems are under version control, you can (probably) get your app up and running at any point of its existence without fuss. You can also see, quite easily, which versions of what gems you were using when. A real history.
I'm working on a project with two other developers that's built on FireBreath. So far, I've been able to get things working perfectly on my machine, but we need to coordinate our development via Mercurial. So I pushed my files to the repository and thought all was well.
Unfortunately, that doesn't work.
The various .vcproj files that make up the solution all contain hard-coded references to my local file system. This works fine for me, because I'm not moving the project around. But when you try to build the solution on another machine with a different file structure (different drive letter, different folder location, etc.) everything breaks.
I used FireBreath's standard project generation script (Python) and then the Visual Studio CMake script (prep2008.cmd) to generate the solution files. What can I do to tweak things so that other developers can use the same code base?
If your developers are not using the same build/make/project files, this could quickly become a maintenance nightmare. So you should definitively all use the same .vcproj files. (An exception to this would be if the project files were generated from some other files. In that case treat those other files in the way described above.)
there's two ways to deal with the problem of differing setups on different machines. One is to make all paths relative to the project's path. The other is to use environment variables to refer to files/tools/libraries/whatever. IME it's best to use relative paths for everything that can be checked out with the project, and use environment variables for the rest. Add a script that checks for the existence of all necessary environment variable, pointing out the meaning of any missing ones, and run this as a build prerequisite, so whoever tries to get a new build machine up and running gets hints at what to do.
To make sure that everyone caught the updated comments from sbi's answer, let me give you the "definitive" answer from the FireBreath devs.
Your build directory is disposable; you should never share .vcproj files. Instead, you should regenerate your build/ directory any time you change the project and on each new computer, just like any project that uses CMake.
For more information, see http://colonelpanic.net/2010/11/firebreath-tips-working-with-source-control/
For reference, I am the primary author of FireBreath and I wrote the article.
I'm not familiar with FireBreath, but you need to make the references relative, and then recreate that relative structure on every machine. That is, if your project sits in "c:\myprojects\thisproject" and has an additional include directory "c:\mydir\mylib\include", then the latter path needs to be replaced with "....\mydir\mylib\include".
EDIT: I rewrote my anyswer to make it clearer. When I got you correctly, your problem is that FireBreath generates those .vcproj files with absolute paths in it, and you want to use this .vcproj files on a different developer machine.
I see 3 options:
Live with it. That means, make sure, every team member has the same file structure / view to the file system, tools installed in the same place.
Ask the authors of FireBreath to change their .vcproj generator to allow relative paths, use of environment variables etc.
If 1 or 2 does not work, write a program or script for changing the absolute path to relatives in those .vcproj files. Run this script whenever you have to regenerate your FireBreath project.
What you should not do due to the FireBreath FAQ: don't change the .vcproj manually, those changes will be lost next time the project is regenerated.
EDIT: seems that "option 4." turned out to be the best solution: generating those .vcproj files for each developer individually. Hope my suggestions were helpful, either.
It would be nice to have a more or less complete list over what files and/or directories that shouldn't (in most cases) be under source control. What do you think should be excluded?
Suggestion so far:
In general
Config files with sensitive information (passwords, private keys etc.)
Thumbs.db, .DS_Store and desktop.ini
Editor backups: *~ (emacs)
Generated files (for instance DoxyGen output)
C#
bin\*
obj\*
*.exe
Visual Studio
*.suo
*.ncb
*.user
*.aps
*.cachefile
*.backup
_UpgradeReport_Files
Java
*.class
Eclipse
I don't know, and this is what I'm looking for right now :-)
Python
*.pyc
Temporary files
- .*.sw?
- *~
Anything that is generated. Binary, bytecode, code/documents generated from XML.
From my commenters, exclude:
Anything generated by the build, including code documentations (doxygen, javadoc, pydoc, etc.)
But include:
3rd party libraries that you don't have the source for OR don't build.
FWIW, at my work for a very large project, we have the following under ClearCase:
All original code
Qt source AND built debug/release
(Terribly outdated) specs
We do not have built modules for our software. A complete binary is distributed every couple weeks with the latest updates.
OS specific files, generated by their file browsers such as
Thumbs.db and .DS_Store
Some other Visual Studio typical files/folders are
*.cachefile
*.backup
_UpgradeReport_Files
My tortoise global ignore pattern for example looks like this
bin obj *.suo *.user *.cachefile *.backup _UpgradeReport_Files
files that get built should not be checked in
I would approach the problem a different way; what things should be included in source control? You should only source control those files that:
( need revision history OR are created outside of your build but are part of the build, install, or media ) AND
can't be generated by the build process you control AND
are common to all users that build the product (no user config)
The list includes things like:
source files
make, project, and solution files
other build tool configuration files (not user related)
3rd party libraries
pre-built files that go on the media like PDFs & documents
documentation
images, videos, sounds
description files like WSDL, XSL
Sometimes a build output can be a build input. For example, an obfuscation rename file may be an output and an input to keep the same renaming scheme. In this case, use the checked-in file as the build input and put the output in a different file. After the build, check out the input file and copy the output file into it and check it in.
The problem with using an exclusion list is that you will never know all the right exclusions and might end up source controlling something that shouldn't be source controlled.
Like Corey D has said anything that is generated, specifically anything that is generated by the build process and development environment are good candidates. For instance:
Binaries and installers
Bytecode and archives
Documents generated from XML and code
Code generated by templates and code generators
IDE settings files
Backup files generated by your IDE or editor
Some exceptions to the above could be:
Images and video
Third party libraries
Team specific IDE settings files
Take third party libraries, if you need to ship or your build depends on a third party library it wouldn't be unreasonable to put it under source control, especially if you don't have the source. Also consider some source control systems aren't very efficient at storing binary blobs and you probably will not be able to take advantage of the systems diff tools for those files.
Paul also makes a great comment about generated files and you should check out his answer:
Basically, if you can't reasonably
expect a developer to have the exact
version of the exact tool they need,
there is a case for putting the
generated files in version control.
With all that being said ultimately you'll need to consider what you put under source control on a case by case basis. Defining a hard list of what and what not to put under it will only work for some and only probably for so long. And of course the more files you add to source control the longer it will take to update your working copy.
Anything that can be generated by the IDE, build process or binary executable process.
An exception:
4 or 5 different answers have said that generated files should not go under source control. Thats not quite true.
Files generated by specialist tools may belong in source control, especially if particular versions of those tools are necessary.
Examples:
parsers generated by bison/yacc/antlr,
autotools files such as configure or Makefile.in, created by autoconf, automake, libtool etc,
translation or localization files,
files may be generated by expensive tools, and it might be cheaper to only install them on a few machines.
Basically, if you can't reasonably expect a developer to have the exact version of the exact tool they need, there is a case for putting the generated files in version control.
This exception is discussed by the svn guys in their best practices talk.
Temp files from editors.
.*.sw?
*~
etc.
desktop.ini is another windows file I've seen sneak in.
Config files that contain passwords or any other sensitive information.
Actual config files such a web.config in asp.net because people can have different settings. Usually the way I handle this is by having a web.config.template that is on SVN. People get it, make the changes they want and rename it as web.config.
Aside from this and what you said, be careful of sensitive files containing passwords (for instance).
Avoid all the annoying files generated by Windows (thumb) or Mac OS (.ds_store)
*.bak produced by WinMerge.
additionally:
Visual Studio
*.ncb
The best way I've found to think about it is as follows:
Pretend you've got a brand-new, store-bought computer. You install the OS and updates; you install all your development tools including the source control client; you create an empty directory to be the root of your local sources; you do a "get latest" or whatever your source control system calls it to fetch out clean copies of the release you want to build; you then run the build (fetched from source control), and everything builds.
This thought process tells you why certain files have to be in source control: all of those necessary for the build to work on a clean system. This includes .designer.cs files, the outputs of T4 templates, and any other artifact that the build will not create.
Temp files, config for anything other than global development and sensitive information
Things that don't go into source control come in 3 classes
Things totally unrelated to the project (obviously)
Things that can be found on installation media, and are never changed (eg: 3rd-party APIs).
Things that can be mechanically generated, via your build process, from things that are in source control (or from things in class 2).
Whatever the language :
cache files
generally, imported files should not either (like images uploaded by users, on a web application)
temporary files ; even the ones generated by your OS (like thumbs.db under windows) or IDE
config files with passwords ? Depends on who has access to the repository
And for those who don't know about it : svn:ignore is great!
If you have a runtime environment for your code (e.g. dependency libraries, specific compiler versions etc.) do not put the packages into the source control. My approach is brutal, but effective. I commit a makefile, whose role is to downloads (via wget) the stuff, unpack it, and build my runtime environment.
I have a particular .c file that does not go in source control.
The rule is nothing in source control that is generated during the build process.
The only known exception is if a tool requires an older version of itself to build (bootstrap problem). In that case you will need a known good bootstrap copy in source control so you can build from blank.
Going out on a limb here, but I believe that if you use task lists in Visual Studio, they are kept in the .suo file. This may not be a reason to keep them in source control, but it is a reason to keep a backup somewhere, just in case...
A lot of time has passed since this question was asked, and I think a lot of the answers, while relevant, don't have hard details on .gitignore on a per language or IDE level.
Github came out with a very useful, community collaborated list of .gitignore files for all sorts of projects and IDEs that is worth taking a look.
Here's a link to that git repo: https://github.com/github/gitignore
To answer the question, here are the related examples for:
C# -> see Visual Studio
Visual Studio
Java
Eclipse
Python
There are also OS-specific .gitignore files. Following:
Windows
OS X
Linux
So, assuming you're running Windows and using Eclipse, you can just concatenate Eclipse.gitignore and Windows.gitignore to a .gitignore file in the top level directory of your project. Very nifty stuff.
Don't forget to add the .gitignore to your repo and commit it!
Chances are, your IDE already handles this for you. Visual Studio does anyway.
And for the .gitignore files, If you see any files or patterns missing in a particular .gitignore, you can open a PR on that file with the proposed change. Take a look at the commit and pull request trackers for ideas.
I am always using www.gitignore.io to generate a proper one .ignore file.
Opinion: everything can be in source control, if you need to, unless it brings significant repository overhead such as frequently changing or large blobs.
3rd party binaries, hard-to-generate (in terms of time) generated files to speed up your deployment process, all are ok.
The main purpose of source control is to match one coherent system state to a revision number. If it would be possible, I'd freeze the entire universe with the code - build tools and the target operating system.
Which of the hg plug-ins has:
the least hassles
causes the least trouble
is prettiest
Can't claim to have tried a wide variety, but, what's wrong with hg4idea...?
In regards to the "don't use one" response - this is hardly adequate. What if I use my IDE to do a refactoring that renames a file? Without IDE/source control integration, the file rename is made without regards to source control, and then Mercurial (or whatever else) thinks a file went missing and a new one appeared. Then you have to go back to wrangle with the source control to sort things out.
JetBrains seems to have chosen hg4idea-luciad for its upcoming Python editor (PyCharm) and it is now more active than hg4idea
It looks like a leader is on the way :-)
To answer your question: The best IDE PlugIn is don't use on.
I think IDE integration is not necessary when working with a DVCS. When working in a centralized System, it is reasonable for the purpose of automatic check out on edit. However, I like keeping things separate. I don't want my IDE cluttered up. I don't see any benefits in using a plug-in compared to a standalone solution (that I keep running on a second monitor etc. ).
I am fine with TortoiseHG and the command line for more complicated tasks.
I want to give credit to all open source libraries we use in our (commercial) application. I thought of showing a HTML page in our about dialog. Our build process uses ant and the third party libs are committed in svn.
What do you think is the best way of generating the HTML-Page?
Hard code the HTML-Page?
Switch dependency-management to apache-ivy and write some ant task to generate the html
Use maven-ant-tasks and write some ant task to generate the HTML
Use maven only to handle the dependencies and the HTML once, download them and commit them. The rest is done by the unchanged ant-scripts
Switch to maven2 (Hey boss, I want to switch to maven, in 1 month the build maybe work again...)
...
What elements should the about-dialog show?
Library name
Version
License
Author
Homepage
Changes made with link to source archive
...
Is there some best-practise-advice? Some good examples (applications having a nice about-dialog showing the dependencies)?
There are two different things you need to consider.
First, you may need to identify the licenses of the third-party code. This is often down with a THIRDPARTYLICENSE file. Sun Microsystems does this a lot. Look in the install directory for OpenOffice.org, for example. There are examples of .txt and .html versions of such files around.
Secondly, you may want to identify your dependencies in the About box in a brief way (and also refer to the file of license information). I would make sure the versions appear in the About box. One thing people want to quickly check for is an indication of whether the copy of your code they have needs to be replaced or updated because one of your library dependencies has a recently-disclosed bug or security vulnerability.
So I guess the other thing you want to include in the about box is a way for people to find your support site and any notices of importance to users of the particular version (whether or not you have a provision in your app for checking on-line for updates).
Ant task seems to be the best way. We do a similar thing in one of our projects. All the open source libraries are present in a specified folder. An Ant task reads the manifest of these libraries, versions and so on and generates an HTML, copies into another specified folder from where it is picked up by the web container.
Generating the page with each build would be wasteful if the libraries are not going to change often. Library versions may change, but the actual libraries don't. Easier to just create a HTML page would be the easiest way out, but that's one more maintenance head ache. Generate it once and include it with the package. The script can always be run again in case some changes are being made to the libraries (updating versions, adding new libraries).