I am in the process of migrating between two operating systems on the same computer. This has got to the stage of being a multiboot system; both OS are Windows Vista. The aim ultimately is to replace the old (and creaking) system with another Vista OS. One is intended as a working system and the other as an experimental system where I can indulge in some of the hairier techniques I pick up on the web and elsewhere without unduly compromising the working system.
I now have data from the old system replicated in a separate disk partition so that it can be accessed from either OS. Aptana Studio 3 is set up on both systems and I have successfully imported projects from one to the other. However, one component missing from the process is the local history files, so that I cannot get back to previous versions of project files from the new installation. I have discovered that these files are located in %WORKSPACE%.metadata.plugins\org.eclipse.core.resources.history, but have not found a way of accessing them from either installation. Copying them across from either the data partition mentioned above or from the old OS doesn't appear to work, and anyway would defeat the object of being being able to access project files, including their histories, from either system.
What I am looking for is a way of migrating the Aptana workspace in the old system, complete not only with projects (which I have done and which works) but also with their local histories so that both Aptana installations can access them on completely equal terms.
I hope that makes sense. Please bear in mind that I am an enthusiastic amateur rather than an IT professional; that may influence the range and depth of any advice that may be offered.
Related
I am developing on multiple machines and have the repository and/or project-folder on a private cloud.
I would like to have a file or something, that includes every used tool (NP++ v1.x.x, VS2019 v4.x.x, yEd v2 etc.).
I find the idea of the "package.json" for NPM extremly useful. Maybe their is something similar for OS-Level. (Win10 by the way)
Possible solutions I thinked of:
of course just track it manually
Virtual Machine (which I don't want to use and cannot host anyway)
The tool/practice/extension/whatever should only track some given IDEs/tools. Not setup a OS from zero.
I have a question about the rules organizing resources into components.
What I want to do is the following; There are two files (both 64bit) with the same filename but in different source locations and with different component GUID's;
C:\data\win7\driver.sys
C:\data\win10\driver.sys
I want both included in the installer and when the installer runs decide depending on the OS which of these two files will be installed in the same location:
C:\Program Files\MyProgram\driver.sys
Is this a violation of the rules regarding the components? (the main one applying would be the first one I would say:)
"Never create two components that install a resource under the same name and target location. If a resource must be duplicated in multiple components, change its name or target location in each component. This rule should be applied across applications, products, product versions, and companies."
https://learn.microsoft.com/en-us/windows/desktop/Msi/organizing-applications-into-components
Actual Answer: Yes, installing different files to the same absolute path is indeed a violation of the MSI component rules - in
theory. As you have discovered reading the documentation. However,
seeing as only one file will make it onto the box, it is all semantics
really, there will never be two files installed. Still, the MSI will
probably not validate - you will see validation errors.
I am not sure of the technical implications, but could you make the
file name different based on OS? Then you simply put the files in
separate components and use different component GUIDs and condition
the component for installation or not. Full compliance instantly.
With all that said, it seems driver files are to be installed using
INF-files in standalone packages in the future. Please read below.
Component Rules, Pragmatically: This is a well-known problem indeed. I have this old answer that might be helpful: Change my component GUID in wix? It describes how the component rules work in practice, and how the idea is that you reference count an absolute path, and not a file per-se. Please read it to make it clearer. More links in this answer.
Driver Installation: We are supposed to know how to do this, but the rules keep changing (and another answer). Essentially drivers are to be distributed via Windows Update, or at least via a standalone package without the need for an installer in the future (until Microsoft change their mind again?). Windows Hardware Dev Center dashboard. As far as I understand the installation is INF-based (as before).
Question: Are you sure about that installation location? What does the documentation say? I thought sys files should go primarily to the Windows folder? Or the WinSxS folders (side-by-side win32 assemblies). As you understand driver installations confuse us all.
Windows 10 Detection: It appears to be a challenge to detect Windows 10 due to the new "evergreen versionlessness" of Microsoft. I have this old answer written up to summarize my current understanding of it: Windows 10 not detecting on installshield. Deployment tools such as Advanced Installer does the job for you and detects Windows 10 with simple measures. Not sure what Installshield does. The linked answer lists a few approaches to detect the version yourself - not sure which is the saner approach. Please read (and also please report any interesting findings when you have a successful package).
Recently I started to incorporate good practices in my development workflow, so I split the development server and the production one. I also incorporated a versioning system using Subversion (Tortoise SVN).
Now I have the problem of synchronize the production server (Apache shared hosting) with the files of the last development version in my local machine.
Before I didn't have this problem because I worked directly with the server files through Filezilla. But now I don't know how to transfer the files in an efficient way and what are the good practices in this aspect.
I read something about Ant and Phing but I'm not sure if this appropiate to me or is unnecessary complexity.
Rsync is a cross-platform tool designed to help in situations like this; I've used it for similar purposes on multiple occasions. This DevShed tutorial may be of some help.
I don't think you want to "authomatize" it, rather establish control over your deployment and integration process. I generally like SVN but it has some bugs and one problem I have with it is that it doesn't support baselining -- instead you need to make a physical branch of your repository if you want to have a stable version to promote to higher environments while continuing to advance the trunk.
Anyway, you should look at continuous integration and Jenkins. This is a rather wide topic to which not a specific answer can be given. There are many ins, outs, what-have-yous. Depends on your application platform, components, do you have database changes, are you dealing with external web services or 3rd party APIs etc.
Maybe out there are more structured solutions but with Tortoise SVN you can export only the files changed between versions in a folder tree structure. And then, upload as always in Filezilla.
Take a look to:
http://verysimple.com/2007/09/06/using-tortoisesvn-to-export-only-newmodified-files/
Using TortoiseSVN, right-click on your working folder and select
“Show Log” from the TortoiseSVN menu.
Click the revision that was last published
Ctrl+Click the HEAD revision (or whatever revision you want to
release) so that both the old and the new revisions are
highlighted.
Right-click on either of the highlighted revisions and select
“Compare revisions.” This will open a dialog window that lists all
new/modified files.
Select all files from this list (Ctrl+a) then right-click on the
highlighted files and select “Export selection to…”
Side note:
You have to open more details about your workflow and configuration - applicable solutions depends from it. I see 4 main nodes in game: Workplace, Repo Server, DEV, PROD, some nodes may be united (1+2, 2+3), may have different set of tools (do you have SSH, Rsync, NFS, Subversion clients on DEV|PROD). All details matter
In any case - Subversion repositories have such thing, as hooks, in your case post-commit hook (executed on Repository Server side after each commit) may be used
If this hook (any code, which can be executed in unattended mode) you can define and implement any rules for performing deploy to any target under any conditions. You must only know
Which transport will be used for transferring files
What is your webspaces on servers (Working Copies of just clean unversioned files - both solution have pro and contra sets) - it will define, which deployment-policy ("export" or "update") you have to implement in hook
Some links to scripts, which export files, affected by revision (or range of revisions) into unversioned tree
Our C++/QT desktop application for Mac, Windows and Linux needs an installer. I'd rather we have a single installer for all three platforms. I do know it's a bit tricky, I guess what I wanted ask is if a framework already exists for that (Java maybe?).
I'd really like to avoid having to write three different installers.
The link that Kyle mentions is pretty comprehensive, but I wanted to provide a bit more of information about InstallBuilder for Qt (Disclaimer, I am one of the developers) since most of the cross platform installation programs referenced there are Java-based. This requires bundling a JRE, etc. and adds a significant overhead that is not required with a Qt-based installer, like ours. It is able to generate wizard-like executable installers for all platforms from a single project file as well as native packages such as DEB and RPM. If you ship your software in DVDs, you can create a single multi-platform DVD that shares data across platforms but still have native launchers.
Having said this, since your application is Desktop-oriented, for the particular case of OS X if it does not require complex installation you may be better off creating a .app file and package it inside a DMG. Users can then drag the file directly to the Applications folder.
Finally, I wanted to mention that InstallBuilder is commercial, but we offer free licenses for open source projects and discounts for small development firms.
IzPack rocks: http://izpack.org. It is truly crossplatform, very lightweight, easy to master, and produces excellent results.
After fully integrating both Izpack and InstallBuilder into our builds (using Windows, OSX, and Ubuntu 14 build servers for testing purposes), I will say I believe InstallBuilder is well worth the money (and free for open source projects according to wojciechka).
Izpack is a bit slow, a bit large if you need to package a JVM, and has an amateurish user interface. Version 5 (release candidate 3) was also not generating uninstallers properly. That said, as long as you use a 4.x version and require a JVM anyway, it may be enough for your needs. The extension interface is not terribly easy to deal with, but is almost infinitely flexible. The Windows installers do not register with the Control Panel uninstaller list.
InstallBuilder has a great, fast UI in the produced installers and has a serviceable UI for creating installers. The XML is pretty easy to deal with, too. Downloads are about as small as possible. It also includes nice hooks for doing all sorts of custom stuff easily. The only slightly annoying thing I ran into was that the Windows server required that I manually add some configuration to set executable bits on the other systems' packages (other systems didn't require this configuration).
Note: I was using the three-platform version of standard InstallBuilder (not InstallBuilder for QT).
I am to migrate a Websphere machine (including the applications which run on it) to a new machine. They wanted a clean install of the OS and WebSphere, so I did that. I also took a full file backup of all of the applications they had on the old server. The problem is that to re-install them on the new server, the WebSphere dialog asks me for the JAR/EAR/WAR file, which I don't have.
Is there any reasonably easy way to simply extract the backup of the WebSphere application files I have taken from the old maching, and simply configure the new machine to use them? WAR, etc. is a nice feature to have, but to be forced to use it seems silly.
Edit: The existing WebSphere server is still up and running in production.
Edit: The old server is WAS 3.5, which means it doesn't even have an export function, sadly. Also, the directory where it actually runs the content from has a completely different structure (consisting of like a a %/Web and %/Servlet, where % is the context path of the application). In the "Install" section, it doesn't even mention EAR or WAR, only JAR. I am currently thinking that perhaps the best thing to do might be to just copy the directory over to another WAS 3.5 system and then upgrade that system (and hope it converts the folder structure and updated the config as part of the upgrade).
Edit: The closest thing I have found to a solution so far is this link:
http://www.javazoom.net/services/newsletter/was4.html (though I am not sure if that tool is available or relevant for WAS 7.x).
This has to be a problem other people have run into before, but I can't find a solution anywhere on the WEB.
Thank you!
Here do they have sample Jacl scripts one can use to export/import appserver's configuration. So that is what you can start with. If your new bow uses the same version of WAS (and the same topology if it is not a standalone box) as the old one, it might be a (relatively) safe process.
Migration between different versions of Websphere might be somewhat more tricky, but I'm sure IBM published at least one redbook on that topic.
If you still have the old server running, than just export the apps and you have the war/ear files. However, If you don't know the configuration for the apps, you are screwed. However, I am sure IBM has tools that you can use. Some of the paid tools look even nice and user friendly (at least according to their sales demos). I can't tell you what you need, since I don't know what documentation you have for your apps. But as it looks like there is not much there, otherwise you would just install the application the same way they were installed on your old server and use the binaries (war, ear, jar) that are archived somewhere.