[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Git mirrors

From: Stephen J. Turnbull
Subject: Re: Git mirrors
Date: Mon, 17 Oct 2011 20:57:51 +0900

Eli Zaretskii writes:

 >   http://netsplit.com/2009/02/17/git-sucks/

Have you read the comments?  Every third one seems to point out that
git usability is almost as low as Emacsen!  Boy, git is really in bad
company there! :-)

 > Try disregarding its obvious exaggeration and disgust, and just _read_
 > the portions of the man pages reproduced there.

Sure (I don't need to, I've read them myself), but that's almost
irrelevant to introducing such a tool into the general workflow of a
project.  It has to be, because very few tools (even ones with a Daddy
Warbuck$$ like Canonical behind them) come with Emacs-quality docs.
Eg, bzr as you point out. ;-)

In practice I find this is not a huge problem for a free software
project's official VCS.  There will be a ${VCS}For${Project}Devs
document, and there will be people around who are able to walk
new/occasional devs through the workflow with the new tool.

You'll note that many of the people bitching in that blog are people
whose companies forced them to use git, or for whom participating in
the kernel workflow means using git.  I feel sorry for them because
(a) they're probably too busy to read the tutorials and (b) their
colleagues are too busy to answer questions.  Uh-oh, I see great pain
in their future....

 > I often find myself in a similar conundrum, even though I never
 > needed to do something as complex as publish a branch.

Publishing a branch with git is a real wart, granted (at least it most
certainly used to be).  But that's not a doc problem, it's a design bug.

I cannot sympathize with the "'Ref' say what???" issue, though; pretty
clearly the author wants git to present itself the same way that other
VCSes present themselves, and apparently didn't read the tutorial, so
doesn't know the first thing about git.  ("What is a 'ref'?" is indeed
the *first* thing about git -- it untangles almost *everything* in
git, including "repository corruption", which is usually a name for
"oops, I didn't keep a ref to that commit and now I can't find it").

 > [git people having trouble adjusting to bzr] could well be out of
 > habit, though.

It is most definitely not habit in my case.  It's lack of colocated
branches, and lack of technical documentation for pipelines and looms.
I need technical documentation because I have a very clear idea of
what workflows I want to implement, and the use cases and tutorials
published for those tools don't match my workflow.  Sadly it seems
pipelines can't do it (they're different from Mercurial queues) and
looms I've never figured out.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]