|
Branching is evil and should be avoided. It's a sign of a flawed process.
You'll never get very far if all you do is follow instructions.
|
|
|
|
|
PIEBALDconsult wrote: Branching is evil and should be avoided. It's a sign of a flawed process.
Why? I'm on the fence (having seen the usefulness of it in Git), but want to know your reasons.
Marc
|
|
|
|
|
Marc Clifton wrote: but want to know your reasons.
Best I can remember he claims that every production delivery is perfect so he never needs to make an emergency fix into a production release.
|
|
|
|
|
PIEBALDconsult wrote: Branching is evil and should be avoided. It's a sign of a flawed process.
The fact that you don't like it and that you claim that you deliver perfect code into production doesn't mean that the rest of the world doesn't need it nor that they shouldn't have it.
|
|
|
|
|
Well, they want back to the first feature of source control : backup !
I definitely need to write that article about it...
~RaGE();
I think words like 'destiny' are a way of trying to find order where none exists. - Christian Graus
Entropy isn't what it used to.
|
|
|
|
|
Not sure what you are stating but just to be clear - source control is not a back up system.
|
|
|
|
|
Hear hear!
You'll never get very far if all you do is follow instructions.
|
|
|
|
|
Preach it brother and all the rest who have posted!
I've had something happen with "that thing" that none of the "experts"/acolytes where I work could explain why it happened.
In our repository there are many directories. I was working on some files in directory A(name changed for its protection and simplicity). One of my coworkers was working on files in directory H. For the record, I didn't even know about directory H. Our files were completely disjoint. I go to commit my files by creating the commit, marking which files I want to commit, entering the commit label, etc. While doing this I see I am behind by one commit so I dutifully do the Pull. The coworker's files come down to directory H. Git then creates, on its own mind you, a second Push commit for me to merge the files I just pulled, the ones from directory H which I knew nothing about until the pull, back into the repository.
I saw this one other time but the behavior seems to be random. Yes, I have contacted GitHub and they are looking at it.
|
|
|
|
|
I never got into Git because the name is so stupid.
I'm tired of stupid names for technology.
I'm relieved that it sucks - now I can allow my curiosity die and think of it no more.
|
|
|
|
|
MehGerbil wrote: I never got into Git because the name is so stupid.
... says MehGerbil... 
~RaGE();
I think words like 'destiny' are a way of trying to find order where none exists. - Christian Graus
Entropy isn't what it used to.
|
|
|
|
|
*moves Rage over to the 'not nice' column*
|
|
|
|
|
Never tried it. No plans to try it. Using TFS at work and at least it's a step up from Subversion.
You'll never get very far if all you do is follow instructions.
|
|
|
|
|
Er, no.
Regards,
Rob Philpott.
|
|
|
|
|
I cannot agree with that. I've been using TFS for the past 6 mths, and while I found it to be great Application Development Lifecycle tool, when it comes to the source control, SVN is far superior IMHO.
Initially I thought it is just a matter of getting used to the tool and I should give it a chance. After months of experience, I still reckon that common things like branching, merging, blaming, comparing revisions (changesets) or even committing is simpler and more intuitive in SVN.
|
|
|
|
|
Powerful tool. With power comes responsibility.
~RaGE();
I think words like 'destiny' are a way of trying to find order where none exists. - Christian Graus
Entropy isn't what it used to.
|
|
|
|
|
I could say the same thing about a pneumatic drill, but it doesn't mean I want to use one!
|
|
|
|
|
No, but if you do, you'd better understand how it works...
Decentralised configuration management systems are not trivial (centralised neither, but state-of-the-art now). Plus if you do not need the decentralized way, you may use it as a plain normal centralized version as well.
~RaGE();
I think words like 'destiny' are a way of trying to find order where none exists. - Christian Graus
Entropy isn't what it used to.
|
|
|
|
|
Rage wrote: Decentralised configuration management systems...
Presumably that was a misstatement.
Rage wrote: Plus if you do not need the decentralized way, you may use it as a plain normal centralized version as well.
Doesn't alter the fact that is missing a primary feature for anything above a small company - that of management of multiple products where there are non-trivial code dependencies between them.
|
|
|
|
|
jschell wrote: Presumably that was a misstatement.
How that ? No, dfinitely Distributed/Decentralised Configuration Management System.
jschell wrote: here there are non-trivial code dependencies between them
Define "non trivial" ? DCMS can do everything what a normal CMS can do, so either what you are trying to achieve is not feasible with centralised configuration management, or your code structure was not made up properly.
~RaGE();
I think words like 'destiny' are a way of trying to find order where none exists. - Christian Graus
Entropy isn't what it used to.
|
|
|
|
|
Rage wrote: No, dfinitely Distributed/Decentralised Configuration Management System.
Because that isn't what Git is.
http://en.wikipedia.org/wiki/Configuration_management#Overview[^]
Rage wrote: Define "non trivial" ? DCMS can do everything what a normal CMS...
If we are talking about Git then it does not have a mechanism for dealing with independent deliverables which share code (not other deliverables) because each repository is optimized for dealing with a single deliverable. This works well for open source internet projects. It doesn't work for a company with different product lines because the end up kludging solutions either with multiple repositories or a single repository.
Other source control systems do.
|
|
|
|
|
jschell wrote: Because that isn't what Git is
Source control is one part of configuration management. To be precise, since this is what you are after, let's call it a distributed version control system instead of configuration management system, even if one can argue a version control system in software development can handle about everything required by configuration management.
jschell wrote: If we are talking about Git then it does not have a mechanism for dealing with independent deliverables which share code
Still do not understand what Git cannot do. How would you do that with Subversions ? Or ClearCase ? or SourceSafe ? or Vault ?
~RaGE();
I think words like 'destiny' are a way of trying to find order where none exists. - Christian Graus
Entropy isn't what it used to.
|
|
|
|
|
Rage wrote: Source control is one part of configuration management
Yes - so git is one part of configuration management.
Rage wrote: Still do not understand what Git cannot do. How would you do that with Subversions
I can manage (check out, version, etc) only one directory under a tree of directories in Subversion.
Git requires multiple repositories for the same ability.
|
|
|
|
|
Sure. Let's drive a Panzer VI instead of a car. More power, more responsibility. Heil Git.
|
|
|
|
|
But can it get from Berlin to Warsaw on one tank of fuel?
You'll never get very far if all you do is follow instructions.
|
|
|
|
|
Allow me to bring my modest contribution.
(CEF build excerpt)
$ git svn find-rev r251746
_
Oh, I forgot to say: this was started 2 hours ago. Thanks, I have my Ctrl/C.
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
