I ran into my first Core Data versioning problem - learn something every day!
Following instructions found here, I made a new version of the model, added the code for lightweight migration, and then went to set the active version…
Uhhh, where do you do that? The docs don't actually say, and other threads here talk about "click on the main file". WHAT "main file"?
The original xcdatamodel has no version number in it. Is that a problem? Is the Migration Manager still going to be able to figure this out?
All I did was add a field, this seems like a lot of work…
Core Data model files don't use version numbers. The files might include a number in their name, but that's for people to see, Core Data doesn't care about it. It uses entity hashes to compare models.
The "main file" is the .xcdatamodeld that contains all the versions (which have names ending in .xcdatamodel).
Select that then look in the file inspector pane on the right. It has a pop-up menu that you use to select the current version.
Related
I'm trying to generate NSManagedModels from my datamodel. Generation works but after I got many errors :
error: filename "Station+CoreDataProperties.swift" used twice:
'/Users/Me/MyApp/Models/CoreData/Station+CoreDataProperties.swift' and
'/Users/Me/Library/Developer/Xcode/DerivedData/MyApp-gwacspwrsnabomertjnqfbuhjvwc/Build/Intermediates/MyApp.build/Debug-iphoneos/MyApp.build/DerivedSources/CoreDataGenerated/Model/Station+CoreDataProperties.swift'
:0: note: filenames are used to distinguish private
declarations with the same name
I try clean build folder and derivedData directory hard delete. I'm using Xcode 8 BETA maybe it's a bug ?
I get this in Xcode 8.1
For me following steps solved the issue. Please note that order matters.
1) Create entity in Core Data model.
2) Under class section, make settings as on following image.
Module: Current Product Name
Codegen: Manual/None
3) Generate your NSManagedObject subclass.
This post greatly helped me solve this problem myself. Personally I look at this as an Xcode bug. Bug or not this is a huge chicken and egg situation.
I ran into this by:
Created a new Project using Core Data
Generated my NSManagedObject subclass+extension (while codegen: ClassDefinition)
I accidentally saved the generated classes in the Wrong folder
I deleted the generated files
Re-generated in folder I wanted
💥- Xcode used twice errors
As others have posted I kept cleaning my build (and clean build folder) but that never fixed the build issue.
I finally figured out if you originally created your NSManagedObject generated classes with codegen: ClassDefinition, as I did without knowing then you are locked in for the chicken and egg issue.
I then deleted the auto generated classes thinking I had to re-generate, so I did. Once re-generated I would get the used twice build error again. I manually went into the ../DerivedSources/CoreDataGenerated/Model/.. and deleted the duplicates. Again, I re-generated thinking I'd only have 1 copy (in my project) but I was wrong. If codegen: ClassDefinition was originally set then Xcode will keep creating the auto-generated classes+extensions and put them in the buried folder ../DerivedSources/CoreDataGenerated/Model/... I repeated this chicken and egg a few times before catching on.
I later realized you do indeed need to mark codegen: Manual/None however to get things back in sync you need to delete the auto-generated files in ../DerivedSources/CoreDataGenerated/Model/.. and in your project if you have any there still.
Be careful setting codegen: Manual/None, for me it was bit tricky because codegen: Manual/None wouldn't stick. I had to click back and forth between entities multiple times to double/triple check each entity was set to codegen: Manual/None. Then auto generate the files. At this point your only copy of the auto generated files should be in your project and not in ../DerivedSources/CoreDataGenerated/Model/...
Last, I think this is a bug because if you specify codegen: Manual/None I don't expect Xcode to auto generate files at all, yet it does and puts them in your project. More confusing if your setting is codegen: ClassDefinition, who the heck knows Xcode will put the files in a buried directory yet it is available for use in your project. My beef with this is the auto generated files aren't source controlled and if I change computer I have to know to auto-generate them on the new station.
Hope this helps someone else!
Cheers!
This is indeed not a bug. As #Morrowless suggests both class definition and properties extension are created. If this is not wanted, select Manual/None under Codegen before generating the code. If the code is already generated, just delete them, and try Editor->Create NSManagedObject Subclass... again from the menu (after setting Manual/None).
Note, in the picture below, the Class Name 'Contact' is specific to my project. You will see your entity name instead.
If you generated CoreData subclasses with codegen: ClassDefinition your basically screwed. The only way to fix it is to:
Delete your CoreData subclasses.
Delete your derived data folder.
Clean your project (CMD+K).
Generate new CoreData subclasses, this time select Codegen: Manual/None and Module: Current Product Module
This is not a bug. Codegen generates these files in the DerivedData folder, so you don't need to create them again in your project, hence the compile error.
From Xcode 8.0 Release notes:
Xcode automatically generates classes or class extensions for the entities and properties in a Core Data data model. Automatic code generation is enabled and disabled on an entity by entity basis, and is enabled for all entities in new models that use the Xcode 8 file format. This feature is available for any data model that has been upgraded to the Xcode 8 format. You specify whether Xcode generates Swift or Objective-C code for a data model using the data model’s file inspector.
When automatic code generation is enabled for an entity, Xcode creates
either a class or class extension for the entity as specified in the
entity's inspector: the specified class name is used and the sources
are placed in the project’s Derived Data. For both Swift and
Objective-C, these classes are directly usable from the project’s
code. For Objective-C, an additional header file is created for all
generated entities in your model. The header file name conforms to the
naming convention “DataModelName+CoreDataModel.h”.
However, if you selected Category/Extension under the codegen pulldown menu in the data model inspector (because you want to add logic to your model): codegen will wrongly generate both the class definition and properties extension.
The solution is to simply delete the properties extension (ClassName+CoreDataProperties.swift). Your project should now compile.
After following the guidance from oyalhi and Vladimir Shutyuk, (deleting the NSManagedObject files, changing the entity codegen to Manual/None), I had to restart Xcode to allow it to index again before I could re-generate the NSManagedObject files and get a successful compile.
For the sake of completeness..:
I just ran into the same error, but none of the proposed solutions worked. What puzzled me was that even switching from automated code generation to manual for the one (as I thought) problematic entity didn't do anything.
Finally, I figured out that I had several entities with the same name, but they all shared the same classname. The reason for this was that I copy&pasted one entity several times to save me some work, because they also have a few attributes in common.
Turns out XCode renames the duplicates by adding 1, 2,... to the entity name, but leaves the class name as before. And since now entity name and class name are "unrelated", renaming the entity won't change the class name either.
Hope it helps someone - I have also filed a bug report for this.
I'm working on an app and need to make an update to the xcdatamodel file.
So I created a new file:
I have model.xcdatamodel
and now I have model2.xcdatamodel
However, it looks like the app is still using model.xcdatamodel. How can I change that to point to model2.xcdatamodel?
I see that next to model.xcdatamodel, there is a "green check icon".
Thank you,
Tee
Ah I got it. Just need to select model.xcdatamodeld (i.e the parent of all xcdatamodel files) and change the "Versioned core data model" to the one you want to use.
I have been using SubSonic 2 on several projects before but with the new SubSonic 3 I have implemented in 2 projects. However, my question has always been is if I can change the the output T4 template to generate a class file for each table instead of single ActiveRecord.cs file. I want to use it in a very large project and I can see where is not practical to have 80+ tables in a single file. I prefer to have separate class files.
Would I need to change SubSonic.Core?
If its not possible, please let me know.
Thanks
Why does it matter how many files there are if the code is entirely generated? What practical difference is there?
You can change the templates to output multiple files. No changes would be required to the SubSonic dll, just the T4 Templates.
However, I fail to see how it is worth even just the time to post the question here, much less the time required to actually make those changes.
There is a way to do this, if you rewrite the T4s to follow this example. However, I think there is an issue that may arise when you drop a table, the previously created .cs file for that table will not be removed. I think you would have to further edit the T4 to start by deleting all its previously generated files.
We have a lot of VBA code in spreadsheets and a lot of time people save them to local drives. When we want to upgrade the spreadsheets we push a new version out to a shared drive but dont have any way of enforcing that people dont use the old versions of the spreadsheets.
Is there some best practice here to deploy vba spreadsheets so if someone loads an old version it wont open or will ask you to upgrade. It seems like this must be an issue for any custom solution so i would have through MS would have some solution here.
Does microsoft have a standard versioning / deployment solution for this or do i need to come up with some home grown solution (spreadsheet pings a database on startup to check version)
If its just the VBA code you want to be able to upgrade one solution is to separate all the code into an XLA that sits on an accessible server.
then the users Excel has an addin loader that gets the latest version from the server, or you could embed the loader in the workbook.
There is a working example reversioning Addin Loader available at
http://www.decisionmodels.com/downloads.htm
Though I've not done this. I've see people do something similar where they store the code in dlls in the database, and then verify that the local ones are the same on start up.
Add a custom property to the workbook, such as a GUID, Integer, Date, or whatever you need. On startup, check the value and determine if VBA should continue or whatever implementation you decide. Usually, I open the "shared" latest version of the document and inspect its custom property. If user is not using latest version, display a message prompting the user to get the latest version. This is usually good enough. I have done this for years with great success for Access, Word, and Excel VBA.
One real problem is people who ignore the prompt to update. They will do this because they fear losing data, bugs in latest version, and so on. You need to address their concerns and not try to "lock" them in to your "solution" to this problem. I strongly recommend you always provid a means of "importing/upgrading" data to latest version of workbook.
Fairly simple to implement. If you address the above mentioned problem effectively, people will start to trust the "prompt" and you will find this is a really simple and effective solution.
If you workbook is tied to a database, it gets a little more complicated. Generally, you do not allow the user to maintain the data outside of the database. When they want to modify the data, you generate the workbook for them. After they are done modifying data, you import the data. (The workbook is saved for "backup" purposes or the user can maintain a copy of the workbook for reference or an archive.) This has the advantage of eliminating the need to maintain document version since the document is "virtual".
Define a current version property. Use whatever way for that, like a constant in VBA.
Find a standard way to access the latest version. It could be on a shared drive always on the same path, or whatever other means you have.
If current version number < latest version number then:
Ask to upgrade
Show the upgrade procedure
Possibly do some locking on normal VBA operations until the upgrade is done
An easy solution would be:
Store a version number somwhere in the workbook (on a hidden sheet for example so that it will not get removed accidentally) or as a variable.
Store a text file with the latest version number on your share
Include a macro which automatically reads the latest-version-textfile and compares the version number to the one in the currently running workbook. If its too old just show a message-box (telling the user to update) and close the workbook.
I'm using a similar approach and it works fine.
I made a discovery some time back. Just follow these steps:
Create a .doc/.xls/.ppt file in office 2003. Keep some test data in there and close the file. Now rename the file to change it's file extension to a random string, taking care that it is unassociated, like test.asdfghjkl etc.
Double click the file and it opens seamlessly in the parent application.
Now AFAIK, windows checks the file extension of the file and uses it to do an action, viz open an application and pass the file to it to open. Then how does the office suite manage to do this?
EDIT: How about the case when the extension is changed to one that is associated with another application. Is there a priority algorithm in place for handling that ?
Do you have the "View extensions for known types" option on?
EDIT: #Comments....
Yes, its a stupid/insulting question, but when troubleshooting a problem I have learned to assume nothing, and trust the users 0%.
BUT, I tried it, and you're right. Its stupid that MS has this kind of behavior, and it can only lead to security vulnerabilities, which led me on a search for your answer.
From the posts at http://seclists.org/fulldisclosure/2007/Jan/0444.html
"You have stumbled on an age-old
quirky behavior of Windows. Office
document formats are based on a
standard Windows container format, OLE
structured storage files, also known
as "docfiles". A docfile's name and
extension are irrelevant - the file
is, conceptually, a serialization of
an OLE object, and like all
serialization formats it contains the
identifier of the application that
produced it, in the form of an OLE
class id (in GUID format) in this
case. You can easily verify that it
doesn't work with the newer Office XML
formats"
Indeed it doesnt work for the 2007 *X file types, but 2K3 is still a problem. To solve this problem... Upgrade! =)
And here at security focus under TOC point 2.
So, there you go.
I can't seem to make this happen now, but I know I saw Windows reading XML processing instructions a few years back. Maybe that is what's going on?