Here is my situation :
I have created a homebrew tap to host some tools, let's call it me/poor-name.
Now I would like to transition to a better named tap called me/tools for various reasons.
Is there a way to print some kind of deprecation warning for a while when my users install tools from me/poor-name, so that they progressively use me/tools as a tap ?
Or is there a better way to handle the transition ?
Homebrew doesn't support renaming taps, but it does allow migrating formulae from one tap to another.
Create a pull request to the new tap adding the formula file as-is from the original tap. Fix any test failures that may occur due to the stricter requirements for new formulae than existing formula (e.g. brew audit --strict must pass for that formula).
Create a pull request to the original tap deleting the formula file and add it to tap_migrations.json with a commit message like gv: migrate to homebrew/core.
Put a link for each pull request in the other pull request so the maintainers can merge them both at once.
For example, here is the tap_migrations.json file for brewsci/science: https://github.com/brewsci/homebrew-science/blob/master/tap_migrations.json
Also, this feature should work the same with casks. (It's just not documented.)
Related
I want to know how go get works, because when im using a libary from github (for example gorilla/mux) the documentation said to install it I should type "go get -u github.com/gorilla/mux", it strange without specific version.!
As far as i know it always use from master branch.
What happen if the owner of the library update the code? Is it gonna be deprecated or something?
Because in my case, im using docker to build the project, every time I build the image, it always do "go get"
Is there any best practice to using go get?
You may use go get library#version. For example, go get foo/bar#v1.2.3
I have created a Basic MSI Installer using InstallShield 2014 for a server/client program and have to hide features dynamically based on the License Key of a database that is installed prior to our Server app being installed. I have created conditions for the features that need to be hidden, setting the InstallLevel to 0 if they are not licensed and 1 if they are licensed. I am getting the license key after the SQL Login dialog (because the installer wouldn't know what database to look in otherwise) but conditions are evaluated during the CostFinalize action, which runs before the dialogs are created. So after I get the license key and run some other custom actions to determine the availability of each feature, I call the CostFinalize action before the CustomSetup dialog is shown.
I am getting the correct behavior for the features that need to be shown, and you can select or deselect said features in the dialog, however, when the installation executes, the selected feature is not installed....and the log file says that the feature is not selected for install, even though the user clearly selects it. Why would this be happening? Is there another approach to hiding features dynamically (I have tried the FeatureSetData function in an InstallScript action, but to no avail)?
Also, after I added the conditions to the features, whenever I try to uninstall the program from the Programs and Features app, I get an Error 1606 Could not access network location. It's like the registry key gets messed up when there are conditions on the features...Any help would be greatly appreciated.
I found the problem...If you set the features InstallLevel to 0 to start with and have a condition that sets it to something greater than zero, then it will not install the feature, regardless of whether it's selected. If you invert this logic and start with the features InstallLevel set to 1 and have a condition that changes the installlevel to 0, it will hide or show the feature AND it will be installed properly. This also caused the error 1606 I was getting on the uninstall...
Also, if anyone ever has components that get installed that aren't supposed to be installed, then you might try switching the Dependency Checking to none. For some reason, the .NET dependency check that InstallShield does causes certain components to install all the time, even if their assigned feature is turned off. Hope this helps someone in the future.
The CostFinalize can also be run by a dialog to refresh the feature list. Here are the steps:
In the Next PushButton of the SetupType Dialog, create a new item at the top.
Event: DoAction
Argument: CostFinalize
Condition: 1=1
In my case, I was hiding a feature based on a previous dialog and needed it to reevaluate conditions in the Program Feature.
Condition: Level:0 GLOBAL_VAR=0
Condition: Level:1 GLOBAL_VAR=1
Up until last week, I had been using RubyMine for my Rails development. I know it has a vim plugin but I have been working on migrating my development to vim and tmux. I don't want to keep using the mouse and VIM gives me a lot more flexibility. I have found plugins and workarounds for almost all the features I care about except the "interpreted auto complete" functionality in my first screenshot below. RubyMine interprets the whole rails application and offers sorted-by-relevance suggestions (as you can see, it's showing me instance variables and methods for the class in question and the modules it includes) THEN it shows (less relevant) methods available on the Object class. It also shows the method signature when there's one.
Also, in my second screenshot, you can see how RubyMine offers autocompletion for core Ruby classes.
Compare this to the bottommost screenshot. I do have completion but there's no way to find what I'm looking for. I'm using ctags , YouCompleteMe, vim-rails, vim-ruby and I also tried installing eclim to see if it makes a difference.
Is there a plugin I've missed that can enhance my auto completion? It doesn't look like RubyMine is doing something super crazy. pry can give me the same 'power' if it were running in the same 'context'.
First Screenshot (RubyMine interpreted auto complete):
Second Screenshot (RubyMine core Ruby classes auto complete):
Third Screenshot (vim omnifunc + ctags):
Important Note
This solution only works for Ruby 1.9+
I forked 'vim-ruby' at https://github.com/zxiest/vim-ruby and modified it as such:
Method signatures now appear in completion.
I disabled sorting by name for methods.
Plugins and Settings
vim-rails
I'm using vim-rails https://github.com/tpope/vim-rails
supertab
I'm using supertab https://github.com/ervandew/supertab instead of YouCompleteMe (mentioned in my question) although YouCompleteMe is super fast and automatic but there are currently some compatibility issues between my it and my vim-ruby fork.
vim-easytags
I'm using vim-easytags https://github.com/xolox/vim-easytags
Add this to your ~/.vimrc
:set tags=./tags;
:let g:easytags_dynamic_files = 1
Make sure to touch ./tags in your project directory.
Issue :UpdateTags -R **/*.* from vim in order for easytags to generate your tags file.
Remap omnicomplete
In order for omnicomplete to pop up, by default, we have to hit <C-X><C-O>. I remapped this to <C-Space> by inserting the following in my ~/.vimrc:
inoremap <C-#> <C-x><C-o>
I now press tab when I want supertab to complete my code and Ctrl+Space when I want omnicomplete to trigger and show method signatures for me. There's definitely a better way to integrate this (i.e. getting supertab to call omnicomplete after a dot)
And here goes the screenshot! Notice that method sorting being off allowed my custom resize method to appear on top and the signatures now appear in the completion (as well as in the editor when enter is pressed!)
just tried to create a snapshot of one of my projects. It didn't work, and Xcode did throw the following error at me: "Unable to create a snapshot. fatal: You don't exist. Go away!"
Source control commit doesn't work either. Ok, obviously there's something wrong with my local repository. I don't have the slightest idea, what that could possibly be, though. I have not committed in quite a while, but except some updates to Xcode itself, nothing major happened to the project folder, as far as I can remember.
So, what can I do about it? Is it possible to deactivate source control for that project and then start over again with source control with the projects current status? I don't need the previous versions anymore, so that doesn't need to be taken care of.
Thank you!
Looking into the git source code, this error message is generated only if
pw = getpwuid(getuid());
fails.
It indicates that the system wasn't able to find an entry for your user account in /etc/passwd or equivalent.
This can happen if your account is removed (via deluser, userdel, or something else) while you're logged in, or if you don't have read permissions on /etc/passwd (the latter should never happen).
Try the following commands at a terminal prompt:
whoami
id
ls -l /etc/passwd
grep "^${USER}:" /etc/passwd
According to the git documentation, this is the meaning of your particular error:
You don't exist. Go away!
The passwd(5) gecos field couldn't be read
This typically means the system doesn't know who you are (you'll see this sort of error on a Linux system if you delete someone's user account while they're still logged in). Is everything else working? That is, can you log out and log back in without a problem? Do other terminal commands operate correctly? What about the id command?
Can you interact with your repository on the command line? What happens if you cd into the directory and try something like git status?
WendiKidd's solution of just starting from scratch is probably the simplest thing to do, assuming it works, but you'll lose any change history already associated with your project.
I had this issue with git and running
dscacheutil -flushcache
in the terminal fixed it.
I would suggest creating a new project entirely, copying your code etc. files out of the old project, and simply migrating over into a new one (which you could then recommit to another source control directory). That seems like the easiest solution; this is a very odd error, and I've come across things in the past that are either unfixable or would take more effort to fix than taking 10-15 minutes to just set up a new project.
So that's what I'd recommend--hopefully that will fix your problem, and whatever xcode is mad at is in some hidden file or the .xcodeproj itself, and not in the files you'd need to migrate over.
My setup project installs a windows service. Right now this action is in the Install category. Would the commit category be a better location for it?
I've added a custom action to remove the service in the Uninstall category, but when I run the uninstall, the user is prompted to shut down the service. As the user should never really know that the service exists, I'd rather this never pops up, but rather stops the service on its own and simply uninstalls this. I can't seem to find a setting for that. How might I accomplish it?
Should I have anything in the rollback section? I added the same custom action for the install/uninstall, figuring it would be smart enough to uninstall the service if the rollback occurs after the service was installed. Is this correct? Would it be moot if I installed on commit instead?