I'm modifying a build process and I need to do a complete comparison of the contents of two .ear files. That means recursively comparing each archive in the .ear. These .ear files have archives that contain archives.
I've looked at Beyond Compare and Archive Analyzer, but they only do one level at a time. I have to manually drill down into each archive. I'm looking for something more automatic.
Eclipse and UltraCompare do a binary comparison of the two .ears which is not what I want.
How can I compare two .ear files recursively?
zipdiff provides a very good open source solution.
My problem turned out to be more than just expanding the .ear file recursively (I wrote a Java class to do that - recursion made it simple.) Once the .ear files are expanded I have to diff the directories to check for any changes. If anything other than timestamps changed then I know that the build is producing a different binary.
The second problem is that our build process generates hundreds of .xml files and subsequent builds re-generate those .xml files with the elements in different order. I'm not sure why. When I expand two .ear files made by back-to-back builds with no changes to anything the diff of the resulting directories shows hundreds of .xml files with diffs, even though they are functionally equivalent.
In addition to expanding the .ear files recursively I need to do a diff and exclude the .xml files in certain directories. I thought that Cygwin diff would do this, but the --exclude switch doesn't recognize any path information:
Cygwin diff won't exclude files if a directory is included in the pattern
If I don't find a solution to this I'll write another Java class to step through the whole directory structure doing a single level diff in each directory and excluding the .xml files in the appropriate directories.
I have the feeling that I'm re-inventing the wheel, but I can't find a wheel right now.
In Beyond Compare go into the Session Settings dialog, and on the Handling tab is an Archive Handling option. If it's set to As folders always BC will treat archives just like folders, so it's fully recursive.
Related
Every time I simulate in Dymola, a number of "useless" (for me) files are created in the working directory - i.e. dsfinal.txt, dsin.txt, dslog.txt, dsmodel.c, dymosim.exe. I find it annoying as it messes up my directory.
Is there a way to select only the desired output files to be kept after the simulations, without the need of manually deleting the undesired ones?
Those are temporary, but necessary files for Dymola. As far as I know there is no option to delete them automatically. Of course you could script that, but I don't see a real point to it and those files are used by some functionality - e.g. dsfinal.txt is used when as simulation is continued.
Some notes: Those files are created in the working directory - which should contain temporary files only. The working directory can be set via the GUI using File -> Options -> Settings:
A rather common problem is, that there is a Open and a Load function in Dymola:
As the description states, Load does not influence the working directory, whereas Open sets it to the directory from which a file is opened. The latter is also true for opening files e.g. via a double-click from the explorer. So usually it is better to go with Load.
My advice would be to separate the directories in which models/packages are stored and the working directory. This way the working directories content can be fully deleted basically anytime...
I have temporary files created in separate folders inside my source tree which I would like to ignore. Something like:
project/
|--component/
|--.jazzignore
|--file.src
|--file-9df29e29373e66caef72/
|--file.src.tmp
I already ignore file.src.tmp by extension using .jazzignore, but I would also like to ignore the file-9df29e29373e66caef72/. The folder looks empty in the "Unresolved" category for the component, but since its name changes over time, I cannot ignore it by name.
since its name changes over time, I cannot ignore it by name.
Still, if you know its naming convention, you might consider an ignore pattern:
core.ignore= \
file-*
Note it is non-recursive, you that would ignore any file, folder or symlink named file-... anywhere under component.
Here, that would ignore only file-... directly under component.
Eclipse workspaces often include files or folders, such as compiler output, log files, and so on, that you do not want to place under source control.
You can specify resources or classes of resources to be ignored by Rational Team Concert™ source control. Ignored resources are never checked in.
A .jazzignore file is used to prevent items from being checked into change sets.
A .jazzignore file consists of a series of patterns. Any file, folder, or symbolic link whose name matches a pattern cannot be committed to a change set.
There are two types of patterns in a .jazzignore file:
core.ignore patterns, that are effective in the same directory as the ignore file; and
core.ignore.recursive patterns that affect items in all of the directories below the .jazzignore file.
How do I set up cscope in Emacs when my source code lives in multiple directories?
Say I have several paths for my C++ project:
/path/to/my/code (and subdirectories)
/path/to/other/code (and subdirectories)
/path/to/static/linking/include/files (and subdirectories)
/path/to/static/linking/lib/files' (and subdirectories)
I would like to use xcscope to navigate/look up symbols in my code and the library that I am linking to.
The instructions for xcscope.el say that I should first run C-c s (Cscope->Create list and index) at /path/to/my/code, but I am then confused about how I to have the other paths indexed by cscope.
The documentation says I should go to the other directories and run cscope -b, but what I am supposed to do after that?
I looked at the cscope.files file that C-c s built. I think I am supposed to add my other paths to this file, but this file includes a list of source code files (not directories).
Do I have to manually edit cscope.files to add every single file that I want to index that is outside of my project's root directory?
The documentation is confusing.
I got one sol'n from http://cscope.sourceforge.net/large_projects.html. Still confusing.
cscope.files is aptly named. It is NOT directories. It is files, only files. Which is as daunting as it sounds. What if you have hundreds of files. Then you have to write hundreds of lines, one per file, into cscope.files. No joke.
Automate it with some scripting. E.g., on linux, use bash scripts and redirection (>,>>,|) operators to select and filter files from any and all directories into that cscope.files.
What about the directory where cscope.files resides? IF you don't include those files, it'll only find them when you open your project (in emacs, but probably applicable to any IDE) in that source directory. So, include them, too, for easy opening of your project from anywhere.
The option "recursive/-R" seems obvious to send to cscope-indexer. Nope, not with many root directories. Does nothing (probably could do something with scripts or elisp or who knows). Just feed those files, absolute path, into cscope.files. And you might have to make sure you have only one cscope.files per project. Don't split them up relative to each diretory. Or maybe you can, something to look into.
Remember, include the /path/to/each/file/ before the file if you want cscope.files to point to it from anywhere. cscope isn't "smart", it's dumb, it just takes directions for where to look and it won't know where to look for "filename", without knowing where it is. You're just asking it to call locate filename, which it can't do anyway, when you give it a lonely filename without a dir path outside of the directory that a particular cscope.files resides.
I hope there's a way to use xcscope inside emacs, just adding directories which it will catalog and index, as the xcscope docs and emacs menu suggest. But I didn't succeed in making it work that way.
Wouldn't a soft link (ln -s) work? Worked for me.
Is it possible to exclude files from debugging in Eclipse?
There are my files which I have written and I am interesting in going through these files using debugger Step Over command.
But there are also many imported library files and Step Over goes into these files, which I'd like to avoid. So is there an option eother to specify which files to debug or which files to exclude?
If you're using a JVM based language (as you didn't specify), under Preferences, filter for Step Filtering.
You will see a list of packages. Add your packages to the list, and ensure Use Step Filters is selected.
Assuming it's an entire package you want filtered out, and you're using Java, this should do what you want.
I'm doing some testing to ensure that the all in one zip file that i created using a script file will produce the same output as the content of a few zip files that i must manually click and create via web interface. Therefore the zip will have different folder structure.
Of course i can manually extracted them out and using my powerful eyeball technique to scan them or even lazier i can write a script to do that, but before i invest more time and get accused by my boss for company time robbery, i'm asking if there's a better way to do this?
I'm using perl LAMP stack by the way.
thanks.
You can use perl's Archive::ZIP or Python's zipfile to extract the filenames, sizes and CRC checksums of the files in the archives. Create a file which contains the results sorted by file name (ignore the path).
For your smaller ZIPs, merge the results of the script (cat list1 list2 list3 | sort).
Now, you can use diff to compare the results.
I can wholeheartly recommend Beyond Compare. Unless you're really getting underpaid, it's the biggest bang for your (bosses) buck.
[Edit] I seem to have scanned over the different folder structure, sorry about that.Beyond Compare can compare all files in folders with the same folderstructure. It does not have (I believe) the intelligence to go searching for matches in files in different folders.
Regards,
Lieven
Create a crc checksum for your files.
If your checksum is the same for the original files and the unzipped files, you can be sure the files are the same. And even works for non text data.
A checksum be easily be created with an external program such as "SFV Checker" or programmatically (.net/java for example include libraries to do this).
Taking a cue from Carra's answer...if A.zip is your single big archive and B.zip is the archive generated through the web then use the following algorithm
Extract all files from A.zip and recursively (w.r.t folders) compute the checksum of the files present in the folder (using cksum, md5sum etc) where the contents were extracted and save this information after sorting it (pipe it through sort) to a file (say A.txt)
Do the same for B.zip and generate B.txt
Compare A.txt with B.txt they should be exactly the same.
OR
Use unzip -l to get file/directory lists for both the (zip) archives and then flatten the hierarchy of the user generated zip file and compare with the contents of your script generated zip file using some thing like diff. By flattening of hierarchy I mean you may need to do some kind of pre-precessing on one or both lists before you can do a meaningful comparison with diff.