On Mar 13, 2015, at 6:40 AM, Graeme Pietersz <[email protected]> wrote: > > My reply to this turned out to be rather long and a bit of rant, so I turned > it into a blog post.
Some critiques: 1. The post is visually divided into two sections, pro-Fossil and pro-Git, but the actual prose frequently intermixes positives of both. You should separate the two. Cover all the things that are great about Fossil in the first section, then explain in the second why you choose to use Git despite Fossil’s advantages. The biggest example of this is your 2-point numbered list in the first section. If it were my post, I’d just fold these two points into the “one big advantage” section following the pro-Fossil section. If you’d rather discuss tools and services in a separate section, that’s fine, but as it is, your post feels more like a brain dump than reasoned exposition. 2. Linux is not “the largest software project ever.” Not even close. Windows’ code base is ~3 times larger. Visual Studio and MS Office are each about twice as large as Linux. The highly-successful Alcatel-Lucent 5ESS telephone switching system was last reported to clock in at over 100 MSLOC. [1] Healthcare.gov was reported at 500 MSLOC, making it about 28x bigger! [2] 3. I rarely use Git, and never speak more than pidgin Git to it, so perhaps I am just ignorant about some fact about the Git GUI situation, but I can’t see why you ding Gitk but then later in the post praise Gitg. It feels like you’re going out of your way to find bad things about Git which have been fixed elsewhere. Is there a good reason not to just ignore the existence of Gitk? I’m all for Fossil boosting, but I don’t see a reason to manufacture negatives for Git. It already has enough genuine negatives to sink it, in my book. :) 4. Regarding email integration, I don’t blame Fossil for not tackling this one. The hosted VCS services have an advantage in that their sysadmins work out how to connect up to the public email system. Even at its best — as seen in popular GUI email programs — an email server connection can be difficult to set up. To even get that far, you have to write a whole lot of automated probing code to work out which security features some particular email server has, whether the username has an @ sign in it or not, whether it wants incoming email on port 25, 587, 465, 475, or 2525, etc. Email is a hot mess. My point is that you’re kind of mixing apples and oranges here, as well as in other areas. Your comparison should separate Git vs Fossil from ChiselApp/self-hosting vs Github. Yes, I know, ChiselApp still doesn’t compare to Github, so Git will continue to come out on top in that comparison. However, this is still the only fair comparison. Also, I believe that once you extract the hosting services from the comparison, Fossil comes out quite a bit ahead of Git. 5. You’re missing one of Fossil’s advantages: truly private project hosting, as opposed to “We think it’s private, but there hasn’t been a publicized hack on Github yet” semi-private hosting. A lot of organizations are going to be unwilling to pay for “private” hosting at Github or BitBucket for that reason alone. That’s true here, for one. What exists in the Git world to compare to Fossil as a private, self-contained, all-in-one service? 6. There are cases where Fossil’s single-executable philosophy really matters. The ones I’ve run into amounted to cross-compiling: it’s easier to build an ARM executable for a Chromebook or Raspberry Pi and copy just that across than to set up a whole cross-compilation toolchain complete with shared libraries, package managers, etc., then ship some massive bolus-of-code over to the other platform and unpack it there. You don’t always get the luxury of building on the target platform. ChromeOS doesn’t even come with compilers. [1] That’s probably significantly outdated by now. [2] Of course, it probably collapsed under its own weight, but still... _______________________________________________ fossil-users mailing list [email protected] http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users
