Hi Carlos, Sean and list -
I'm picking up on this thread rather late (looking around for info
on ... using CVS in a web-shop). I'm just getting beginning to get my
head around CVS, and also going to read Sean's useful document. If
people are interested I'll post up our workings on this area.
Part of the confusion with websites it seems to me, is that there is
very little source & object code distinction, the main distinctions
being in *binary* files like photoshop/fla flash --> gif-png-jpg/swf
flash.
Binary files
------------
Does CVS handle binary files too? the notion of "source code control"
seems to preclude binaries... I will also do a search on the list for
using CVS with JSP, because the complexities of JSP->class->java
seems interesting as well.
My current ideal (before considering implementing CVS):
-------------------------------------------------------
I had been considering a system (manually managed! ouch!) for
versioncontrolling with a version such as v1.2 r12 p4, where v1.2
would be the client-facing version (major/minor changes to site -
major being total design/purpose change - minor being feature
addition, not just content change) and r12 would be a review number
(each time client reviews site the number is incremented. Only some
of these are then uploaded to a live site). the p4 is a production
version number - freely changable by the internal project team to
freeze.
* As a new feature is added it would start as FEATURENAME.r1.p1 and
then get incremented until it was merged into the source tree and
becomes v1.3.rX.pY.
* As a new design of the site is added it would start as
NEWDESIGNNAME.r1.p1 (or as 2.0.r1.p1) and when uploaded to the live
site would be merged in as 2.0.rX.pY).
* The key here is that it is feasable that all the following could be
going on at once:
- ongoing content changes to the site (within 1.2 // timescale 1 wk)
- new features to the existing site (within
FEATURENAME=1.3 //timescale = 1 month)
- new site design/logic/purpose (within NEWDESIGNNAME = 2.0 //
timescale = 3mo)
(a) can you see my logic here? am I making sense?
(b) how much of this would one want to/could one implement in CVS -
would these simply be multiple branches stemming off the tree and
then reintegrating?
Relevant Apache tricks:
-----------------------
It also may interest people to know the apache mod-rewrite / vhosts
alias tricks for hosting multiple sites automatically:
If you want to have a tree that looks like:
serverroot/
websitename/
versionid/
html/
cgi-bin/
source/ (not for website)
and want to maintain root level links (esp to /cgi-bin/ etc)
then you can use mod-rewrite (with pain) or vhosts-alias to map from:
http://versionid.websitename.servername/
without need for separate apache config stanzas for new
websites/versionnames.
I only mention because it took some time to work out THAT this was
possible (thanks to magnus bodin www.x42.org).
Best
Tim
--- In [EMAIL PROTECTED], Carlos Costa Portela <[EMAIL PROTECTED]> wrote:
> > > Here at demasiado.com, we're using cvs to help us develop our
> > > website.
....
directory (http://172.16.100.99/~ccosta/).
</snip>
