<?xml version="1.0" encoding="UTF-8"?><!-- generator="wordpress/2.3.3" -->
<rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	>
<channel>
	<title>Comments on: Limbo: Why users are more error-prone with git than other VCSes</title>
	<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/</link>
	<description>Just another GNOME Blogs weblog</description>
	<pubDate>Fri, 29 Aug 2008 03:36:19 +0000</pubDate>
	<generator>http://wordpress.org/?v=2.3.3</generator>
		<item>
		<title>By: Alex Turner</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-224</link>
		<dc:creator>Alex Turner</dc:creator>
		<pubDate>Wed, 12 Dec 2007 16:19:45 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-224</guid>
		<description>I have found similar things using git, but I have found myself getting used to it.  Infact I think that having to explicity add all changes to the index makes you less likely to forget to add a new file, as you are used to doing a git add for changes.  But I agree that it's easy to forget and not commit enough stuff.  The one thing that I absolutely love about git is the ability to go back through your commits and make changes.  I have caused my repository to be uncloneable because I accidently committed a huge log file, and git-pack couldn't pack it in usable RAM.  All I had to do was filter back through the old commits and remove the file from the index, problem fixed.  If I had done this with svn, I would be up the creek without a paddle in a major way.  I have had other reasons to filter old commits to - I added a bunch of data files that I later found I never really needed.  Put 10000 files in my repo without needing them - D'oh, but git lets me get rid of them as if they were never there.  I know this functionality is dangerous, but it's powerful, and I'm really glad of it.  It feels like git was built by developers for developers and has many features that were built because someone said - oh I'd really like it if git go do this, and then someone made it do that.  The other really awesome thing is git-svn.  Wow - I can manage an svn copy with git - totally rad.</description>
		<content:encoded><![CDATA[<p>I have found similar things using git, but I have found myself getting used to it.  Infact I think that having to explicity add all changes to the index makes you less likely to forget to add a new file, as you are used to doing a git add for changes.  But I agree that it&#8217;s easy to forget and not commit enough stuff.  The one thing that I absolutely love about git is the ability to go back through your commits and make changes.  I have caused my repository to be uncloneable because I accidently committed a huge log file, and git-pack couldn&#8217;t pack it in usable RAM.  All I had to do was filter back through the old commits and remove the file from the index, problem fixed.  If I had done this with svn, I would be up the creek without a paddle in a major way.  I have had other reasons to filter old commits to - I added a bunch of data files that I later found I never really needed.  Put 10000 files in my repo without needing them - D&#8217;oh, but git lets me get rid of them as if they were never there.  I know this functionality is dangerous, but it&#8217;s powerful, and I&#8217;m really glad of it.  It feels like git was built by developers for developers and has many features that were built because someone said - oh I&#8217;d really like it if git go do this, and then someone made it do that.  The other really awesome thing is git-svn.  Wow - I can manage an svn copy with git - totally rad.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Andrew</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-221</link>
		<dc:creator>Andrew</dc:creator>
		<pubDate>Tue, 11 Dec 2007 00:05:11 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-221</guid>
		<description>With bzr, you can use the --strict option with "bzr commit", and it will refuse to commit if there are "unknown" files (i.e. files that bzr hasn't been told to ignore or to track) in the tree.

You can even make this your default behaviour by adding this to your bazaar.conf:

[ALIASES]
commit=commit --strict

(Aliases are documented at http://doc.bazaar-vcs.org/latest/en/user-guide/index.html#using-aliases)
 
If you do botch a commit, it's easy to "bzr uncommit" and then redo the commit the way you intended.</description>
		<content:encoded><![CDATA[<p>With bzr, you can use the &#8211;strict option with &#8220;bzr commit&#8221;, and it will refuse to commit if there are &#8220;unknown&#8221; files (i.e. files that bzr hasn&#8217;t been told to ignore or to track) in the tree.</p>
<p>You can even make this your default behaviour by adding this to your bazaar.conf:</p>
<p>[ALIASES]<br />
commit=commit &#8211;strict</p>
<p>(Aliases are documented at <a href="http://doc.bazaar-vcs.org/latest/en/user-guide/index.html#using-aliases" rel="nofollow">http://doc.bazaar-vcs.org/latest/en/user-guide/index.html#using-aliases</a>)</p>
<p>If you do botch a commit, it&#8217;s easy to &#8220;bzr uncommit&#8221; and then redo the commit the way you intended.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Kalle Vahlman</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-220</link>
		<dc:creator>Kalle Vahlman</dc:creator>
		<pubDate>Sun, 09 Dec 2007 08:59:48 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-220</guid>
		<description>You should also add to the end a note that even though it's easy to leave a file /edit from the commit, git makes it very easy to correct that:

  git add foo.c
  git commit # Whoops, forgot bar.c
  git add bar.c
  git commit --amend # re-make the commit with the correct content

Botching a commit is so common (in every VCS) that the ability to re-visit and edit the commit or even the whole history is what really sold git to me.</description>
		<content:encoded><![CDATA[<p>You should also add to the end a note that even though it&#8217;s easy to leave a file /edit from the commit, git makes it very easy to correct that:</p>
<p>  git add foo.c<br />
  git commit # Whoops, forgot bar.c<br />
  git add bar.c<br />
  git commit &#8211;amend # re-make the commit with the correct content</p>
<p>Botching a commit is so common (in every VCS) that the ability to re-visit and edit the commit or even the whole history is what really sold git to me.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: sb</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-219</link>
		<dc:creator>sb</dc:creator>
		<pubDate>Sun, 09 Dec 2007 05:50:24 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-219</guid>
		<description>I disagree that git is more error prone...  With svn, I always found myself accidentally committing stuff I didn't intend to.  Then, once I do that, svn makes it so damned hard to back out a mistake.  With git, I specify exactly what gets committed.  Even if I blow it, with git I end up committing too little...  That's far preferable to committing too much.  And, with Git, if I do completely blow it, it's trivial to fix the last commit.

So, my experience is the polar opposite of yours.</description>
		<content:encoded><![CDATA[<p>I disagree that git is more error prone&#8230;  With svn, I always found myself accidentally committing stuff I didn&#8217;t intend to.  Then, once I do that, svn makes it so damned hard to back out a mistake.  With git, I specify exactly what gets committed.  Even if I blow it, with git I end up committing too little&#8230;  That&#8217;s far preferable to committing too much.  And, with Git, if I do completely blow it, it&#8217;s trivial to fix the last commit.</p>
<p>So, my experience is the polar opposite of yours.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Elijah</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-218</link>
		<dc:creator>Elijah</dc:creator>
		<pubDate>Sat, 08 Dec 2007 22:08:24 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-218</guid>
		<description>Woohoo!  I beat the nasty wordpress beast.  [Nothing to see here in this or the previous comment, move along and read the others.]</description>
		<content:encoded><![CDATA[<p>Woohoo!  I beat the nasty wordpress beast.  [Nothing to see here in this or the previous comment, move along and read the others.]</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Elijah</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-217</link>
		<dc:creator>Elijah</dc:creator>
		<pubDate>Sat, 08 Dec 2007 22:07:46 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-217</guid>
		<description>Ugh, I hate how wordpress marks me as 'newren' instead of as 'Elijah' in my comments on my blog.  Let's see if the configuration option I tried to twiddle has any useful effect...</description>
		<content:encoded><![CDATA[<p>Ugh, I hate how wordpress marks me as &#8216;newren&#8217; instead of as &#8216;Elijah&#8217; in my comments on my blog.  Let&#8217;s see if the configuration option I tried to twiddle has any useful effect&#8230;</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: newren</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-216</link>
		<dc:creator>newren</dc:creator>
		<pubDate>Sat, 08 Dec 2007 22:05:30 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-216</guid>
		<description>Um, two people who didn't bother reading the whole article but commented anyway?  Maybe with a couple ads I could make my blog just like slashdot.  ;-)

I already mentioned the status message given by 'git commit'.  Two problems: (1) The status message doesn't exist for people who use the -F option (or the -m option), and (2) I think lots of users will tend to miss that message.  Even after writing this post a month or so ago and being fully aware of the gotchas, I still accidentally hit some of these gotchas a few times since then even when I was using a plain 'git commit' despite the status message that was right there.  Granted git reset or the --amend option to git commit make it really easy to fix, but these gotchas do increase the learning curve and it's something people need to be aware of while working.</description>
		<content:encoded><![CDATA[<p>Um, two people who didn&#8217;t bother reading the whole article but commented anyway?  Maybe with a couple ads I could make my blog just like slashdot.  <img src='http://blogs.gnome.org/newren/wp-content/mu-plugins/tango-smilies/face-wink.png' alt=';-)' class='wp-smiley' width='16' height='16' /> </p>
<p>I already mentioned the status message given by &#8216;git commit&#8217;.  Two problems: (1) The status message doesn&#8217;t exist for people who use the -F option (or the -m option), and (2) I think lots of users will tend to miss that message.  Even after writing this post a month or so ago and being fully aware of the gotchas, I still accidentally hit some of these gotchas a few times since then even when I was using a plain &#8216;git commit&#8217; despite the status message that was right there.  Granted git reset or the &#8211;amend option to git commit make it really easy to fix, but these gotchas do increase the learning curve and it&#8217;s something people need to be aware of while working.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Anonymous</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-215</link>
		<dc:creator>Anonymous</dc:creator>
		<pubDate>Sat, 08 Dec 2007 21:17:05 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-215</guid>
		<description>I'll second Michel's comment: if you actually read the status message "git commit" gives you, you can never have this problem.  It will always tell you about what you included in the commit and what you did not include in the commit.</description>
		<content:encoded><![CDATA[<p>I&#8217;ll second Michel&#8217;s comment: if you actually read the status message &#8220;git commit&#8221; gives you, you can never have this problem.  It will always tell you about what you included in the commit and what you did not include in the commit.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: newren</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-214</link>
		<dc:creator>newren</dc:creator>
		<pubDate>Sat, 08 Dec 2007 20:20:20 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-214</guid>
		<description>Michel: Um, yes, I mentioned that in the last paragraph of the article -- was it not clear?</description>
		<content:encoded><![CDATA[<p>Michel: Um, yes, I mentioned that in the last paragraph of the article &#8212; was it not clear?</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Michel S.</title>
		<link>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-213</link>
		<dc:creator>Michel S.</dc:creator>
		<pubDate>Sat, 08 Dec 2007 20:15:59 +0000</pubDate>
		<guid>http://blogs.gnome.org/newren/2007/12/08/limbo-why-users-are-more-error-prone-with-git-than-other-vcses/#comment-213</guid>
		<description>When using git, I learn to pay close attention to the summary that's included when git fires up the editor and prompt for the commit message. It quite nicely lists the files that are tracked but have unmarked changes.</description>
		<content:encoded><![CDATA[<p>When using git, I learn to pay close attention to the summary that&#8217;s included when git fires up the editor and prompt for the commit message. It quite nicely lists the files that are tracked but have unmarked changes.</p>
]]></content:encoded>
	</item>
</channel>
</rss>
