On Tue, Mar 02, 2010 at 06:12:30PM +0000, Chris Withers wrote: > Tres Seaver wrote: > > - - The docs are intended primarily for folks who want to install and > > run Zope, rather than hack on it. > > Says who? The last comment I had on those docs was from Marius when he > had to go back to a Zope 2 project and wanted to make it buildout based.
I wanted to upgrade Zope 2 on a local non-profit's server. I found the INSTALL document. It gave me many options, which was confusing. I picked the buildout option since I thought it was the one blessed by the Zope community. I went through some paint getting it to work, and I consider myself to be pretty familiar with buildout. In retrospect I think I should've used virtualenv + mkzopeinstance. I'm +10 for having a single recommended install option documented in INSTALL.txt. Let the other options be relegated to appendices in a separate file, for people who think they want buildout/rpm/whatever. I don't much care if that option is buildout or virtualenv or anything else, as long as it is reasonably simple and works well for most users. Currently I'm feeling +0.9 towards virtualenv over buildout, but this feeling slowly oscillates over time. A while ago I preferred buildout over virtualenv; now I just need to experience a certain number of virtualenv bugs and issues to swing me back. ;-) > > - - zc.buildout is *super* heavyweight compared to virtualenv > > A point of view, I don't happeen to agree, especially for the simple > case of an instance... virtualenv doesn't fit my brain, buildout does. That is a curious observation. I find it much easier to understand what virtualenv does and how it does it: the underlying model is simpler. Of course it has fewer features than buildout, which is why I use buildout for *development*, but I've seen the cost of those features in increased complexity and difficulty of debugging when something goes wrong. Actually, now that I think about it more, a buildout would have one advantage over virtualenv: with a fixed versions.cfg your Zope 2 installs won't start failing when somebody uploads a new-improved-and-incompatible version of zope.component into PyPI. > > - - We have two alternate zc.buildout scenarios (install Zope + run > > mkzopeinstance vs. self-contained environment). > > Yes, I'm much more for the latter, but when I tried to make that "the > only way", someone whined, so I tried to stay neutral... There will always be people who whine :( Marius Gedminas -- http://pov.lt/ -- Zope 3 consulting and development
Description: Digital signature
_______________________________________________ Zope-Dev maillist - Zope-Dev@zope.org https://mail.zope.org/mailman/listinfo/zope-dev ** No cross posts or HTML encoding! ** (Related lists - https://mail.zope.org/mailman/listinfo/zope-announce https://mail.zope.org/mailman/listinfo/zope )