For open source projects, it's sometimes difficult to find clear build instructions. As a user, I sometimes have to figure out a build process for a piece of software, but then have no obvious way to make that information widely available. What are good ways to release that information?
Someone please make this a community wiki.
The first place I tend to look is in a readme file at the root of the source tree.
Another option, of course, is, if there's a documentation page for the project (and shouldn't there be?) you could have a document there that is specifically about building.
Related
We've developed a commercial Qt application with Qt under LGPL. We link dynamically. We also made a few changes to a couple of files in Qt to better suit our needs. I realize that according to LGPL, I must make those changes publicly available.
My question is how is that done?
Do I set a webpage on my domain with the modified source codes available for download?
Do I distribute the modified source code in a folder on the client target machine after installation?
Another
way?
You simply have to make them available on request.
It can be a file on your web server, it could just be emailing them to anyone that asks, you can even supply them on floppy disk and make a charge for it.
Ideally you would also contribute the changes back to Qt if it's a bug fix or feature that could be useful to other people, then it may be incorporated in future editions and youcould then just use the standard build.
ps. The letter of the GPL says that you have to provide everything that is needed to create your modified build - so the full Qt sources and the build files. However for a large complex project like Qt, if you have only modified a couple of files it would be useful to offer a download of just these (or even a diff against the specific Qt version)
According to the licence you need to provide your changes back to the community, not just to your customer.
The easiest way is forking the main Qt repository on http://qt.gitorious.org/qt and commit your changes there.
I'm quite pretty new to the concept of CVS. However, I want to start using CVS and thus need to 'check-in' some scripts. I'm using a UNIX server and I know that CVS is installed, since doing a
cvs -v
Gives me the correct version number installed. Now the problem I have in is finding documentation to use CVS. Is there an online tutorial/FAQ someone can recommend. I've scoured Google for information and all I come across are posts for installing CVS ...
What I'm really looking for our sample commands taking a beginner from scratch like Logging in etc.
The meta-answer to your question is not to use CVS, unless you're participating in a project that's already using it. Even the CVS maintainers, as far as I understand, don't recommend it for new projects, but recommend svn instead. If you're being obliged to use it, then this answer isn't helpful; sorry.
If the decision is up to you, then you have alternatives:
svn is the system which is most similar to CVS (as noted in another answer).
Mercurial is a distributed version control system, but the distributed features aren't hugely important if, as your question vaguely suggests, you're working on your own.
Git has broadly the same model as Mercurial.
There are others (including at least bazaar and darcs), but those are the big three.
All of these are heavily used in both small projects and big ones.
I now tend to recommend Mercurial to people, and that's the one I predominantly use myself. There are holy wars possible about this, but I feel that's the one which has the best tradeoff between flexibility, good design, and usability (there's a longer version of this answer...!)
Update: there's a very good Mercurial introduction by Spolsky, which is well worth reading for rationale and pointers.
Use svn instead, lots of documentation for that.
Hmmm... a quick Google search for cvs tutorial returns this as the second hit:
http://www.linux.ie/articles/tutorials/cvs.php
I've quickly glanced over it, and Chapter 3 (Basic CVS Usage) starts with "Logging In" and seems to come pretty close to what you need. If you have any concrete questions, feel free to ask.
Does anyone know of a good patching program that is free? You know, one that can take a directory with your old program in it and compare it to a directory with your new version, and spit out a patch that is only the difference between the two?
Also, I am looking for something that can patch the entire directory, not just one exe.
EDIT:
Thanks for the answers, but I am looking for an end user patch for product updates. Nothing to do with the source.
There is Binary patch and diff, which is free, Windows port available.
I've never used this but it is free. It might be worth a try:
Patch Maker 1.2.
A list of tools here reveals a few marked as "free".
Dispatcher: will use an updater GUI as well. They have a quick demo video on their site.
The best tool I've seen for this purpose is Visual Patch 3.5 . It provides the same functionality that you've mentioned. Although it needs to be purchased since it's not free software.
If you are using Subversion, SmartSVN can create a patch for directories (recursive and all) for the files changed since the last commit. I understand this is only a partial answer and it's not CLI, but it's a really good tool if you deal with an SVN repo here.
SmartSVN patch http://cdn.beerpla.net.lg1x3.simplecdn.net/for_www/screenshots/smartsvn-patch.gif
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
I'm a college student and at any given time I have 4-5 programs I'm writing in various languages for various classes/projects.
At any given hour of the day I might be in the library, at home, in any of our different computer lab classrooms, etc.
Right now my current modus operandi is at the end of each class period or coding session, I gmail myself the current state of whatever I'm working on with an appropriate subject line (ie, "MIPS Assembly Lab 2, Revision #3").
However, this is becoming cumbersome and I'm looking for other solutions.
Restrictions:
No Thumbdrive. I'm about as absent minded as possible while still somehow managing to function. I'll lose it.
Portable or Web Apps only. I can't install non-portable executables. So, if a tool requires an installation wizard or administration privileges, I can't use it. I can use portable executables however, stored in our network drive space given to each student. So, that might open some possibilities.
I'm looking for some kind of online storage that I can easily download the latest files for my project or update those in the online storage, with as little friction as possible.
I've considered using some free version control repository and trying to find a portable executable or web-tool I can use to integrate with it, but I wonder if it might be overkill. I'm not really looking for keeping a revision history.
I've seen videos of things like dropbox and it seems like it is a step in the right direction.
Any suggestions?
A good solution would be to get an account at some hosting provider that offers shells (eg. Dreamhost) and do your work remotely. That way you always have a consistent environment that you can just ssh into from anywhere.
It's far easier to find a run-anywhere SSH client than a run-anywhere filesharing or revision control system.
www.github.com?
Git binaries should be usable without any installation process (I do not get the 'portable' part there, as you do not mention anything about your work environment).
Or, alternatively, a thumbdrive git repository, altough you said that you do not want to use a thumbdrive.
The most simple solution (if you can share the code with the world), is to create a project on Google code. This gives you a subversion repository plus a wiki to sort your ideas and an issue tracker for your TODO list, too.
Today, I prefer Subversion in your situation for two reasons:
There is a command line standalone client (just a couple of files which need no install) for Windows. Git would need either Cygwin or MinGW and a Unix environment of some kind. Too much hassle.
It's a bit more simple to use than Git. Git asks a paradigm shift from your brain and unless you get that right, Git will feel "weird".
For professional work on large projects, I prefer Git :)
CVS, Subversion and GIT all allow to create the repository on a network share. All of them discourage this because, in the case of a network outage, the repository may become corrupted.
So if you have frequent network outages, this might not be an option but frankly, most networks are pretty stable today. And in my 15 years since I use VCS, I never had one corrupt a repo on a share. Most network file systems will try their very best to commit pending writes, so unless the server completely dies, the data will be saved when the hiccup is over.
But if you're still worried, use git because it allows to restore the main repository with minimal data loss from your local copy (see this question for details).
We use CVSDude, who do CVS and SVN, it's a pay service $6/month = 250M, works really well, although maybe you're after something free?
You can get a free account on drivehq to store up to 1GB of data. Nothing fancy, but if you're looking for some place to put software it might do the trick.
You might like to check out Bespin:
Bespin is a Mozilla Labs experiment
that proposes an open, extensible
web-based framework for code editing.
I use Dreamhost's integrated SVN. They have an interface for setting up repositories and user accounts. I work on a mac which comes with SVN installed so the whole thing for me was completely painless. Couple of clicks, point SVN at my server and I was good to go.
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 9 years ago.
fossil http://www.fossil-scm.org
I found this recently and have started using it for my home projects. I want to hear what other people think of this VCS.
What is missing in my mind, is IDE support. Hopefully it will come, but I use the command line just fine.
My favorite things about fossil: single executable with built in web server wiki and bug tracking. The repository is just one SQLite (http://www.sqlite.org) database file, easy to do backups on. I also like that I can run fossil from and keep the repository on my thumb drive. This means my software development has become completely portable.
Tell me what you think....
Mr. Millikin, if you will take a few moments to review some of the documentation on fossil, I think your objections are addressed there. Storing a repository in an sQLite database is arguably safer than any other approach. See link text for some of the advantages of using a transactional database to store a repository. As for bloat: The entire thing is in a single self-contained executable which seems to disprove that concern.
Full disclosure: I am the author of fossil.
Note that I wrote fossil because no other DVCS met my needs. On the other hand, my needs are not your needs and so only you can judge whether or not fossil is right for you. But I do encourage you to at least have a look at the documentation and try to understand the problem that fossil is trying to solve before you dismiss it.
After having used Fossil for more than a year now on non-trivial development projects, I feel confident enough to weigh in on this topic.
Below is my experience so far. I'm comparing against git and svn at times, simply because I know those SCM's very well and comparing makes it easier for me to get the idea across.
I'm totally in love with this SCM, so it's mostly points on the plus side.
What I like about Fossil:
We have a bunch of machines (win/mac/a number of Linux distros), and the single-executable installation is just as beautiful as it sounds. No dependencies; it just works. Git is a messy pile of files and the dependency hell in Subversion makes it very nasty on some Linux distributions, especially if you must build it yourself.
The default Fossil workflow suits our projects perfectly, and more git'ish workflows are possible when needed.
We've found it extremely robust, even on large projects. I wouldn't expect anything else from the guys who wrote SQLite. No crashes, no corruption, no funny business.
I'm actually very, very happy with performance. Not as fast as git on huge trees, but not much slower either. I make up any lost time by not having to consult the documentation every other command, as is the case with git.
The fact that there's a tried'n'true transactional database behind every operation makes me sleep better at night. Yes, we've been through more than one horrible incident of stale and corrupt Subversion repositories (thankfully, a helpful community helped us fix them.) I can't imagine that happening in Fossil. Even Subversion 1.7.x use SQLite now for metadata storage. (Try turning off power in the midst of a git commit - it'll leave a corrupt repos!)
The integrated issue tracker and wiki are optional, obviously, but very handy as it's always there - no installation required. I wish the issue tracker had some more features though, but hey - it's an SCM.
The built-in server and web gui is simply brilliant and quite configurable through css.
We sometimes need to import to and from git and subversion repositories. This is a no-brainer in Fossil.
Single file repository. No '.svn' directories all over the place.
What I miss in / dislike about Fossil:
Someone please write TortoiseFossil for our non-technical Windows users :)
The community isn't that large yet, so it's probably hard for a lot of people to introduce it in their company. Hopefully this will change, gaining all the benefits of a large community (documentation, more testing of new releases, etc.)
I wish the local web ui had a search feature (including searching for file content).
Fewer merge options than in git (though the Fossil workflow makes merging less likely to occur in the first place.)
I hope everyone gives Fossil a run - the world is a better place with stuff that just works and which you don't need to be a rocket scientist to use.
Fossil is small, simple, yet powerful and robust, reminds me some principles of C Culture. Likable by those who develop independently and still collaborate.
Any great project should start with principles and continue them at its core as it gathers more layers (GUI, extra features).
I am impressed with Fossil and starting to use... take a look at fossil
cheers
I'm landing on this page after an year of the last post, recursive add that has been mentioned here is now taken care of.
Fossil mesmerizes me with simplicity especially after I struggled to get a bug-tracking system to work with mercurial. I need to see how to manage multiple projects, publish the repositories for multi-user access and how to do merging, manage patches etc. I get the feeling that it wont be disappointing going forward.
I'm not interested in using it for source-code version control, but I am interested in a distributed version-controlled personal wiki that I can sync between all the machines I use.
damian,
1/ yes, fossil doesn't support recursive add. However there are some fairly simply workarounds such as
for /r %i in (*.*) do fossil add "%i"
on Windows, and
find . -type f -print0 | xargs -0 fossil add --
on Unix.
2/ I saw the message about malformed manifest when you are adding a file with non-ASCII characters in the filename. The problem was corrected in the last build.
Regards,
Petr
I think fossil is really cool. The most important feature for me was easy installation, and developer friendly defaults. I currently use it to keep track of the local changes of my files. (Our project is hosted in sourceforge and kept track in CVS.) This way I can "commit" locally even if it would otherwise break the project, so smaller changes can be kept track as well.
Fossil is good. It is simple and easy to use. If fossil can provide GUI interface to check in and check out, then it would be better (prefer java gui to archive cross-platform GUI).
The main advantages of Fossil are "open source" and "use SQLite database", so somebody can compile fossil source code to make fossil work on google android platform (mobile and tablet devices).
I am trying your vcs right now.
I like the idea of having all integrated. After all, is all i want when i look for a system like this. I am an active user of Mercurial. And i couldn't find an integration with a issue tracker (I try unsuccessfully to set p Trac with mercurial in the past).
After some test i realize that:
1) "add" command is not recursively, or i can not found in the doc a way to do it
2) i write a bat (i work with windows) to add 750 files and i run it (it took a while). When a run commit it jumps with "manifest malformed"
i think you could address this issues and others making a survey like the Mercurial's one in https://www.mercurial-scm.org/wiki/UserSurvey.
you could write me at dnoseda at gmail
i am interested in you work. keep improve it.
regards
ps.: as an mayor improvement you could add something like gitstat
Perhaps an uneducated knee-jerk reaction, but the idea of storing a repository in a binary blob like an SQLite database terrifies me. I'm also dubious of the benefits of including wikis and bug trackers directly in the VCS -- either they're under-featured compared to full software like Trac, or the VCS is massively bloated compared to Subversion or Bazaar.