As a dev, do you use .gitignore to ignore everything and purposefully include? Or do you just exclude with it? - gitignore

We have an internal discussion going here and we are somewhat torn on the best practice for using .gitignore on projects that contain a lot of files (like a CMS).
Method 1
Method 1 would be to purposefully .gitignore all files that come standard with your build. That would generally start like:
# ignore everything in the root except the "wp-content" directory.
!wp-content/
# ignore everything in the "wp-content" directory, except:
# "mu-plugins", "plugins", "themes" directory
wp-content/*
!wp-content/mu-plugins/
!wp-content/plugins/
!wp-content/themes/
# ignore these plugins
wp-content/plugins/hello.php
# ignore specific themes
wp-content/themes/twenty*/
# ignore node dependency directories
node_modules/
# ignore log files and databases
*.log
*.sql
*.sqlite
Some staff members like this approach since if you create something outside of the standard files, for example like a /build folder, then it would automatically be detected for inclusion. However, writing custom theming and plugins require you to add a few layers to this file to "step in" to the folders you want to keep, and generally, the file is a bit messier to read.
Method 2
Method 2 ignores everything, and then you whitelist what you want in the repo. That would look like
# Ignore everything, but all to descend into subdirectories
*
!*/
# root files
!/.gitignore
!/.htaccess.live
!/favicon.ico
!/robots.txt
# theme
!/wp-content/themes/mytheme/**
/wp-content/themes/mytheme/style.css # Ignore Compiled CSS
/wp-content/themes/mytheme/js # Ignore Compiled JS
# plugins
!/wp-content/plugins/my-plugin/**
# deployment resources
!/build/**
Some staff like this since it's cleaner, you have to purposefully add something (which makes accidental adds harder), and it also in effect shows you your .git folder structure.
What is the best practice? Which method do you enjoy and would you recommend doing one over the other?

The second method is the best practice, when it comes to exlude some folder contents of gitignore rules.
It better reflect the following rule:
It is not possible to re-include a file if a parent directory of that file is excluded.
To exclude files (or all files) from a subfolder of an ignored folder f, you would do:
f/**
!f/**/
!f/a/sub/folder/someFile.txt
Meaning: you need to whitelist folders first, before being able to exclude from gitignore files.
It is clearer, shorter (unless you have a large number of folder to whitelist)
What if it is a Joomla install with a large amount of directories and files?
Or what if a core upgrade adds new files or folders
Don't forget you can have multiple gitignore files, one per folder.
That means you can mix and match both approaches.
And you have:
http://gitignore.io/ (which does blacklist when it comes to Joomla application)
github/gitignore (same approach for Joomla)

The ideal .gitignore file, is the one that does not exist.
For some reason, you're deeply intermingling files you want to track via source control, with files you DON'T want to track.
This, I think, is the source of your sadness.
You are mixing git's intended purpose, which is versioning of programmer-edited files, with deployment, which is intended to get the files where they belong in the correct directories.
Your question is not clear, as to whether you think the Wordpress core files should be versioned. I'm assuming not, since that's how you've set up your .gitignore.
Your question is also not clear, as to whether you are deploying a web site, or shipping plugins as a product. Those are both different use cases, and they require different types of versioning. If this is a deployed web site, you SHOULD be versioning Wordpress along with everything else. If you are shipping a plugin or a theme, then you should have a test suite of plenty of different Wordpress versions to test against.
I think your source control system should be set up, solely to track just the plugins/* and/or themes/* files that go into your distribution. Zipping that folder should give you the plugin asset that your customers download.
To debug your plugins, there should be a deploy step in your IDE that copies each of those tracked files, into a Wordpress install at a location you choose. This permits you to more easily test against different Wordpress versions.
You're reducing workflow problems, to trying to choose a .gitignore. Fix the problem at the root by getting the workflow right.

Related

How to specify buildroot build process variable to be called on make <package>-dirclean

Is there anyway I can just call into a define such as LIBFOO_DIRCLEAN, and just do what was implemented in the define?
Inside HOST_LIBFOO_INSTALL_CMDS, I copy files to the target directory, and would like the 'make package-dirclean' to delete what was copied into the target directory. 'make clean', would obviously do this(any many more), but that is much more than I want to do.
I see the following buildroot variables. LIBFOO_EXTRACT_CMDS, LIBFOO_CONFIGURE_CMDS, LIBFOO_BUILD_CMDS, HOST_LIBFOO_INSTALL_CMDS, LIBFOO_INSTALL_TARGET_CMDS, etc.
make foo-dirclean is a simple tool that just deletes the package build directory. In most cases, when the list of files installed by a package does not change over time (only files content changes) you can simply rebuild the package and the target directory will be rebuilt correctly.
If you want you can implement your own foo-myclean step that implements your own logic. However you must understand deleting files in the target directory is not supported by Buildroot and thus you are on your own.

Exclude files or folders while cfx xpi - Firefox addon sdk

Is there any way to exclude a folder while creating an xpi file using cfx xpi. The documentation doesnt provide any info on this.
I am asking this because in the directory created by cfx init i also have a directory as HTMLTestPages in which i have some test JavaScript since we cannot debug the add-on sdk JavaScript.
Every time i have to create an xpi file i have to remove the folder and again place it. furthermore, if we has this folder on svn this is more problematic.
searching on google shows this bug listed to exclude vim temporary files and in the comments they are also discussing excluding any files starting with .(period). So is there any way to check if this is working in add-on sdk 1.6.1?
The current Add-on SDK version will only add very specific subdirectories to the XPI package, namely data and lib. If your project also has a subdirectory foo or files at the top level with names that the SDK doesn't recognize - these will be simply ignored.
Now if your files are located inside a known subdirectory like data - in this case they will be included in the package. The Add-on SDK has hardcorded rules concerning files and directories that should be excluded, you can see them under python-lib/cuddlefish/util.py:
IGNORED_FILE_PREFIXES = ["."]
IGNORED_FILE_SUFFIXES = ["~", ".swp"]
IGNORED_DIRS = [".git", ".svn", ".hg"]
This means that file names starting with a period or ending with either ~ or .swp will be ignored. But as far as directories goes - only .git, .svn and .hg will be ignored (and this definitely works). If you want an additional directory to be ignored then the only way will be to edit python-lib/cuddlefish/util.py and add it to IGNORED_DIRS list (something that you will unfortunately have to repeat after each SDK update).

How should I handle Sphinx configuration in version control?

I have a problem with my development workflow and Sphinx. I want to keep configuration file for Sphinx in version control so it's easier to manage. This means it's easier to link the file to code updates, etc ... However, the configuration file is stored in /usr/local/etc.
There are two solutions I can think of. Store the file in the repository and move it to the correct folder on deployment or recompile Sphinx to look for the file in my repository. I had a suggestion from someone to use a symlink, but that still requires a change on deployment.
Is there an elegant solution in Sphinx I'm missing?
perhaps have the /usr/local/etc/sphinx.conf file be a script that pulls the actual sphinx config from the file in your repo.
http://sphinxsearch.com/docs/current.html#rel098 scroll down to general and you'll see:
"added scripting (shebang syntax) support to config files (example: #!/usr/bin/php in the first line)"

Step-by-step guide of Mercurial for iPhone projects?

I am looking for a step-by-step Mercurial guide for iPhone projects. Please assume:
hg already installed
the audience is comfortable with command line operations
everything is on OS X
As a newbie, I am particular interested in:
what files should be excluded
how to exclude above files
guideline/suggestion of naming builds
any relevant experience to share with
Please do not discuss how hg is better/worse than any other SCS. This is a how-to question, not a why question.
Thanks!
A more general exclusion tip that applies to all source-control systems is that it's really useful to set up Xcode to use a shared build directory somewhere away from the source code. I prefer to use /Users/Shared/builds/ but it could go anywhere. The nice thing is that you don't need to bother excluding build products from hg, and it makes it convenient to exclude them from backups as well.
You can set this up in Xcode's preferences, under "building".
Files you should exclude are: .DS_Store build *.mode1v3 *.pbxuser. You can also always ignore stuff from other SCSs like .git and .svn. I also ignore .LSOverride which is in some of my projects since I have multiple installs of the developer tools (for SDK betas) and sometimes want to override the standard launch service behavior.
These files can be ignored by listing them in a .hgignore file in the root of the working directory. The .hgignore file must be created manually. It is typically put under version control, so that the settings will propagate to other repositories with push and pull [1].

Can you "ignore" a file in Perforce?

I sometimes use the feature 'Reconcile Offline Work...' found in Perforce's P4V IDE to sync up any files that I have been working on while disconnected from the P4 depot. It launches another window that performs a 'Folder Diff'.
I have files I never want to check in to source control (like ones found in bin folder such as DLLs, code generated output, etc.) Is there a way to filter those files/folders out from appearing as "new" that might be added. They tend to clutter up the list of files that I am actually interested in. Does P4 have the equivalent of Subversion's 'ignore file' feature?
As of version 2012.1, Perforce supports the P4IGNORE environment variable. I updated my answer to this question about ignoring directories with an explanation of how it works. Then I noticed this answer, which is now superfluous I guess.
Assuming you have a client named "CLIENT", a directory named "foo" (located at your project root), and you wish to ignore all .dll files in that directory tree, you can add the following lines to your workspace view to accomplish this:
-//depot/foo/*.dll //CLIENT/foo/*.dll
-//depot/foo/.../*.dll //CLIENT/foo/.../*.dll
The first line removes them from the directory "foo" and the second line removes them from all sub directories. Now, when you 'Reconcile Offline Work...', all the .dll files will be moved into "Excluded Files" folders at the bottom of the folder diff display. They will be out of your way, but can still view and manipulate them if you really need to.
You can also do it another way, which will reduce your "Excluded Files" folder to just one, but you won't be able to manipulate any of the files it contains because the path will be corrupt (but if you just want them out of your way, it doesn't matter).
-//depot/foo.../*.dll //CLIENT/foo.../*.dll
Yes, But.
Perforce version 2012.1 added a feature known as p4ignore, inspired by Git. However the Perforce developers made a change to the behaviour, without justification, that happens to make the feature a lot less useful.
Whilst Git takes rules from all .gitignore files, Perforce doesn't know where to look until you specify a filename in an environment variable P4IGNORE. This freedom is a curse. You can't hack on two repositories that use different names for their ignore files.
Also, Perforce's ignore feature doesn't work out the box. You can set it up easily enough for yourself, but others don't benefit unless they explicitly opt-in. A contributor who hasn't may accidentally commit unwelcome files (eg. a bin folder created by a build script).
Git's ignore feature is great because it works out the box. If the .gitignore files are added to the repository (everyone does this), they'll work out the box for everyone. No-one will accidentally publish their private key.
Amusingly, the Perforce docs shows '.p4ignore' as an example ignore rule, which is backwards! If the rules are useful, they should be shared as part of the repository.
Perforce could still make good on the feature. Choose a convention for the file names, say p4ignore.txt, so the feature works out the box. Drop the P4IGNORE environment variable, it's counterproductive. Edit the docs, to encourage developers to share useful rules. Let users write personal rules in a file in their home folder, as Git does.
If you know anyone at Perforce, please email them this post.
This works as of Perforce 2013.1, the new P4IGNORE mechanism was first added in release, 2012.1, described on the Perforce blog here:
https://www.perforce.com/blog/new-20121-p4ignore
As it's currently described, you set an environment variable "P4IGNORE" to a filename which contains a list of the files to ignore.
So you can check it out to see how you like it.
If you want a solution that will apply to all work-spaces without needing to be copied around, you (or your sysadmin) can refuse submission of those file-types through using lines like the below in the p4 protect table:
write user * * -//.../*.suo
write user * * -//.../*.obj
write user * * -//.../*.ccscc
I remember doing this before, but I don't have the necessary permissions to test this here. Check out Perforce's Sysadmin guide and try it out
Perforce Streams makes ignoring files much easier, as of version 2011.1. According to the documentation, you can ignore certain extensions or certain paths in your directory.
From p4 help stream
Ignored: Optional; a list of file or directory names to be ignored in
client views. For example:
/tmp # ignores files named 'tmp'
/tmp/... # ignores dirs named 'tmp'
.tmp # ignores file names ending in '.tmp'
Lines in the Ignored field may appear in any order. Ignored
names are inherited by child stream client views.
This essentially does what #raven's answer specifies, but is done easier with streams, as it automatically propagates to every work-space using that stream. It also applies to any streams inheriting from the stream in which you specify the ignore types.
You can edit the stream via p4 stream //stream_depot/stream_name or right-clicking the stream in p4v's stream view.
And as #svec noted, the ability to specify ignore files per workspace is coming soon, and is in fact in P4 2012.1 beta.
Will's suggestion of using .p4ignore only seems to work with the WebSphere Studio (P4WSAD) plugin. I just tried it on my local windows box and any files and directories that I listed were not ignored.
Raven's suggestion of modifying your client spec is the correct way under Perforce. Proper organization of your code/data/executables and generated output files will make the process of excluding files from being checked in much easier.
As a more draconian approach, you can always write a submit trigger which will reject submission of change-lists if they contain a certain file or files with a certain extension, etc.
HISTORICAL ANSWER - no longer correct. At the time this was written originally it was true;
You can not write and check in a file that the server will use to make ignore rules; general glob or regexp file pattern ignore in perforce.
Other answers have global server configurations that are global (and not per folder).
The other answers show things that might work for you, if you want one line in your view per folder times number of extensions you want to ignore in that single folder, or that provide this capability in WebSphere Studio plugins only, or provide capability for server administrators, but not available to users.
In short, I find Perforce really weak in this area. While I appreciate that those who use the Eclipse Plugin can use .p4ignore, and I think that's great, it leaves those of us that don't, out in the dark.
UPDATE: See accepted answer for new P4IGNORE capability added mid-2012.
I have found it easiest to reconcile offline work using a BASH script like this one:
#!/bin/bash
# reconcile P4 offline work, assuming P4CLIENT is set
if [ -z "$P4CLIENT" ] ; then echo "P4CLIENT is not set"; exit 1; fi
unset PWD # confuses P4 on Windows/CYGWIN
# delete filew that are no longer present
p4 diff -sd ... | p4 -x - delete
# checkout files that have been changed.
# I don't run this step. Instead I just checkout everything,
# then revert unchanged files before committing.
p4 diff -se ... | pr -x - edit
# Add new files, ignoring subversion info, EMACS backups, log files
# Filter output to see only added files and real errors
find . -type f \
| grep -v -E '(\.svn)|(/build.*/)|(/\.settings)|~|#|(\.log)' \
| p4 -x - add \
| grep -v -E '(currently opened for add)|(existing file)|(already opened for edit)'
I adapted this from this Perforce Knowledge Base article.
I'm looking for a .p4ignore like solution as well (and not one tied to a particular IDE). Thus far, the closest thing I've found is p4delta. It sounds like it will do exactly what the original poster was asking, albeit through another layer of indirection.
http://p4delta.sourceforge.net
Unfortunately, while this does seem to produce the proper list of files, I can't get "p4delta --execute" to work ("Can't modify a frozen string") and the project has not been updated in year. Perhaps others will have better luck.
If you are using the Eclipse Perforce plugin, then the plugin documentation lists several ways to ignore files.