Consider a solution has 2 projects: ProjectA and ProjectB (both are MonoTouch apps) and ProjectBase. ProjectBase contains the whole application, but Main.cs file (with the entry point) is located in ProjectA and ProjectB (which reference ProjectBase). This way, running any of A/B projects will boot up the application from ProjectBase.
Now, I want to override something for ProjectA only (it might be XIB file, image or a .NET class). Is there any way I can setup the solution so that the code and resources, produced by ProjectBase, are merged with the ones from ProjectA/ProjectB and the latter wins?
I found (probably a quirky and kinda-undefined-behavior-driven way) of overriding XIBs: I just put a XIB into ProjectA and ProjectB, name it the same as it was named in ProjectBase and them exclude it from ProjectBase. Although MonoDevelop compiles all items, it seems that the startup project's XIBs get priority, so that I see ProjectA-specific XIBs when I launch ProjectA and ProjectB-specific XIBs when I launch ProjectB. However, I am not sure it is the way it should behave, plus, from what I can see from build log, ALL projects get built yielding resources at the end.
P.S. I'm sorry if this has been asked previously, but I was not able to find the similar question on SO.
I was once trying to do this for a bunch of apps. I would have thought build order would be ProjectBase and then ProjectA, and the content copy system would be the same... Guess this means we are wrong.
You could do a few things.
A) Build your own program to copy resources which are marked for content. Would not be very hard, just need to read the .csproj files. XML parsing is easy enough in .NET. Run this program on post build. Would just have to be careful when doing builds such as to zip or to the device as I am not sure how it handles post-build events.
B) [This is what I did instead] If I expect to also make ProjectC, ProjectD ... ProjectN I instead made a program to generate my program... (Programception).
What it does, has ProjectBase, and ProjectTempalte. You enter your new project name into this program, say, "MyNewProject" and it will create the correct folder structure, write the correct csproj files, and update SLN file. ProjectTemplate has various placeholders in .csproj files like {PLACEHOLDER} which Programception would just go through and find/replace with my project name. Image files (and in your case XIB) are then only kept within ProjectA (B..C..N) unless I do not expect to try and override them in which case they would stay in ProjectBase. This is a lot easier with a XIBless application I would assume. I never use XIB's anyway.
Hope that helps!
Related
I have two questions about cucumber frame work.
1-My .features files are not converting into feature format and showing as plan text file even though it works fine and I can run my TestRunner.
2- I have saved my all TestRunner files into a package Runner but in TestRunner file I have to give full path of my feature file in order to run TestRunner. e.g features= "Features" not working but
features="C:\Users\FourStar1\eclipse-workspace\com.freeTourTest\src\main\java\Features2\Addusers.feature"
works.
Take a careful look at the casing of your directory and package names. It's important you are consistent with these. It makes it easier to spot mistakes. Using lowercase is recommended.
Additionally tests and test resources should be put under src/test/java and src/test/resources respectively. Feature files are resources and will not be copied over target if you put them under java.
I would suggest learning the maven project layout by heart. It makes everything easier then making up your own organization.
We are starting a new project using angular2, typescript and gulp. For the time being our application will consist of two subprojects: a components library (which in the future might be spun off into a separate project) and the app using the component library.
The layouts of the project is going to be something along the lines:
/project_root
/component_library
/src
/library
/components
/services
... etc
/application
/src
/app_name
/components
/services
... etc
The components in application will be using components from the library (but not the other way round)
We would like to have clean (non relative) imports in the app components when importing stuff from the library (we want to avoid ugly imports of the sort '../../../component_library/src/library ...etc' plus, what's more important, we want to be able to move the library code to a separate project without the need to update imports.
There are two possible solutions I see (don't like any of them):
Add a gulp task that would watch the component library and on every change copy the file to node_modules in /project_root
Some sort of simlink? so that we can point /project_root/node_modules to /project_root/component_library/src?
I'm afraid the first solution might not work well with IDE autocomplete in the application (first gulp would need to do the compilation/copying then the IDE would need to pick up the change from node_modules - this looks like something that can be really slow)
The second solution feels hacky - it would need to be repeated by everyone who checks out the code from repo.
What would be the best solution here?
what's more important, we want to be able to move the library code to a separate project without the need to update imports.
Ship your component_library with source and add it as a node_module dependency. Then when someone pulls your code they can add a git remote to node_modules/component_library code and work on the two projects seemlessly.
This is the approach I took with ntypescript.
Probably a pretty fundamental question. When developing a single-project-solution in MSVC++ everything makes a lot of sense, every file is visible to the compiler. When moving to a multiple project solution, none of the interface elements governing project dependence seem to make sense (coming from a .net background.)
For starters it appears setting up project dependencies with that terminology seems to only imply what order the projects are compiled in. There is no more function attached to the definition and addressing objects or methods of one project from another will not work. #including the headers of the dependency doesn't seem to work either, so the dependent project cannot see the files of its dependency. Copying a link of a header from one project to another doesn't work either, where for two projects under the 'Header Files' section, the same file is referenced in both locations. Even with this header references do not work and the files are unconnected. The only way of creating a functional dependency is to add the dependency's paths to the linker/compiler search path of the dependent or worse, simply copying the files of one project to another.
Either I'm missing something or the .NET Visual Studio IDEs have succeeded at what the VC++ IDE has badly failed at.
Edit:
For the sake of asking a specific question, take the following steps for creating a solution in VC++ and I ask if the missing step(s) can be filled in:
Create empty DLL project inside new solution.
Create method in DLL project and compile.
Create new empty executable project.
Create main method in new project.
...
Have main method call function in DLL project.
Preferably fill in the missing step(s) with the most most modern/industry standard/best practice method that best maintains project modularity as intended.
From this I would expect to independently be able to extrapolate and create classes, enumerations etc. in the DLL class and access them all in the executable project, so long as I can find out how this is intended to be done.
In step 2, as per the usual C++ rules, declare the method in the header. Read up on __declspec(DllExport) since the default is that methods are internal to the DLL.
In step 5, as per the usual C++ rules, include the header. Visual C++ needs to know where the header is coming from, so you need to reference the source project. Details vary between Visual Studio versions.
In step 6, you call the method in the normal way. Since it's now declared as __declspec(DllImport), the compiler will leave it to the linker, and the linker will get it from the correct DLL.
I have noticed that lately our xpages application build/compilation time has gone drastically up. It can even take almost whole minute to clean application.
It is quite complex app with some controls (custom controls and java controls) and a lot of legacy code (js, ls, java), but similar app build in pure ecplise with the same amout of java code (which is the only part of this app that changes) cleans in less than 3 seconds ...
I have done a bit of research with rcp debug console and noticed that while cleaning there is some insane traffic that sums to 15000 transactions (and this database has only like 300 design elements that contains code at all!)
The log looks like this repeated over and over again:
...
[1240:0007-03E0] (13586-124 [14561]) OPEN_NOTE(REPC12579BB:0033C2FE-NT00003052,00400000): 0 ms. [48+17446=17494]
[1240:0007-03E0] (13587-124 [14562]) OPEN_NOTE(REPC12579BB:0033C2FE-NT00006C12,00400000): 1 ms. [48+32118=32166]
So my question is:
Does this mean that xpages builder is really badly written or maybe I do not know something?
Are you doing Project - Build or Project - Build All. If the latter and you have multiple NSFs open, then the build will run for all apps. You can close the apps in Package Explorer.
When you say there are 300 design elements, is that the number you see in the Applications Navigator? Bear in mind that's just a drop in the ocean of the actual files being built. Package Explorer will show there are additional Java and xsp-config files for each XPage and Custom Control. These are the Java classes compiled down from the XML markup you see in the XPages and Custom Control design elements. But this is not all that gets built. The server or local database cannot run .java files. It needs to run .class files compiled for the relevant platform. In addition, there are other .class files that need incorporating into the application at compile time, which can be seen by selecting Project - Properties in the Package Explorer view. If you have localization, there are still more files for each language for each XPage / Custom Control. And of course there are files like xsp.properties, an activator .java file and .class file.
The XPage builder may seem badly written because these files need creating. But it is actually very clever, because we don't need to write Java and we don't need to define all the relevant Java files the application is dependent on. We don't need to create a new XSPInput or whatever the Java classes are for the various controls. We can just drag and drop, set the properties from pretty panels which give us drop-downs or Boolean selectors etc. Without those the build time for the app after each save would be quicker, but the development time would be slower.
Its appears after all that there is a solution to build time problem - its the new Domino Designer 9. After upgrading to public beta version the build times dropped drastically (to just few seconds)!
I am using a UI component from J2ME Polish. I found a simple bug in the code, and want to quickly fix it but cannot find any build scripts in the download package, although source-code is included.
The problem cannot be solved by extending the class as it involves members with private access.
How should I go about fixing it? Should I simply compile the class in question with Javac or should try to locate build script in their repository?
There is no binary code for J2MEPolish UI library, because these classes are also preprocessed during your application compilation. This is why there is also no build script for J2MEPolish UI part - the resulting binary, if it compiles, would not make any sense :)
If you are using this UI class directly (there is no Polish magic involved) I would suggest to copy it into your own project, make the changes there and you should be just fine.
But there is a small issue. I'm not 100% sure, but Polish build process should be as follows: first your classes are preprocessed and then Polish classes will be preprocessed. If the UI component, that you copied into your project, will stay in the same java package, it could happen that your changes will be overwritten by buggy Polish implementation. You can check it by running the application or looking at the final java code in build directory.
If the overwriting problem happens, then you have two options - move UI class in your project into different package (I'm 99% sure this should be fine if you use it directly) or you need to modify Polish build process, so that it would preprocess Polish classes first and you could overwrite changes done by it.
/JaanusSiim
I know this question has been answered and accepted but an easier method would be to used their built in property (for any one coming here via the magic search engines)
polish.client.source=/projecthome/j2me-polish-source
which is simply a copy of their sources packaged with the installer. You can copy this into your own source tree and thus have version history of your changes.
While JaanusSiim's method might work I would not recommend it as it becomes confusing having de.enough.** packages in your own source tree I normally create a src for my personal source and a src-j2mepolish for their source files this was it is VERY explicit what I have added to the standard versions.