jwcolby
jwcolby at colbyconsulting.com
Sun Jun 13 10:22:15 CDT 2010
Mark / Shamil, I would love to go to Murcurial, it is really just a matter of time to figure it out. I was hoping that I would get a "this is how it is done" from one of my listmates, and specifically how to migrate my existing source to murcurial. I am not wedded to VisualSVN, I just have it, know it (it is easy) and am using it. BTW Even VisualSVN stamps commits with version numbers, and it is possible (I assume) to check out a specific version number to work on should a later build break something. At the moment Paul and I work on separate workstations. We do things in small chunks and do just keep them local (don't check in) until we test and it all seems to work, then we check in. Just because of the way I approach development, we are building small enough chunks of stuff that we pretty much build, test and commit all in one work session, two at the most. I am not justifying not switching, but it truly hasn't yet caused a problem simply because of the scale that I work on. It also helps that the user is one of two developers so that if I do hit a snag, I can usually just go fix it rather than filing a bug report and waiting for the dev team to go fix it. What is really going on is that VisualSVN works just like every other Source control system. I am used to the paradigm and it was dead simple to set up and start using. There is enough different with Mercurial that when I tried to go there I was spending time trying to figure out how to do source control, and I didn't have the time to spend. I am sold on the concept, and someday when things quiet down, or when my listmates who use it are willing to hold my hand / walk me through it, I will undoubtedly make the switch. BTW, what are the best motorcycles? Triumph? BSA? Norton? I didn't know that you were into motorcycles. John W. Colby www.ColbyConsulting.com Mark Breen wrote: > Hello John, > > Let me say, I am not religious on almost any topics (with the possible > exception of the best motorcycles), so I am not favouring Mecurial > over Subversion religiously. > > But here is the problem. Even if you have one programmer - and you already > have two you have the scenario that you work, work, work and then test a > routine. You think that it is good, and want to preserve that current > version. But you really do not want to check it in to you repository. So > what do you do? One option is quickly xcopy some files (this is surely the > medieval option). Another choice is check in and be-dammed. Third Choice > is copy to notepad and hope that we do not have a power cut. > > Now with two programmers this gets even worse, Paul works all day and does > not check in, in the evening time, he wants to check in for safely, but he > cannot be sure that there are no breaking changes in his code, so he cannot > check in his code or he will break your code. So, result is programmers > world wide are afraid to check in their code until they are finished > working. I know about this problem when I worked on a three person team > that were world wide, so I was never really sure who might check out after I > had checked in untested code. > > In effect I had to operate without SCCS. > > Mecurial systematically solves that problem. They made is the first goal of > the product. > > In a nutshell, when you install Mecurial you install first a local version, > which is your private SCCS. Then later when you hire a guy named Paul, he > also gets a private SCCS on his machine. At that point you install a > central SCCS and from time to time you can merge your code to the central > SCCS. Paul can do the same and Mecurial is built from the ground to assist > in this process. > > It solves the problem I had when I was terrified to check in, but I still > wanted local version control. > > Take 20 minutes and read Joel Spolsky's last ever blog post. After ten > years he chose to make it about Mecurial. I see now that FogCreek also has > launched a paid for product based on Mecurial so he must be committed. > > thanks and good luck, > > Mark