[pmwiki-users] RFC: PITS 00701 -- WikiFarm confusion

DaveG pmwiki at solidgone.com
Wed Mar 15 12:35:11 CST 2006


I agree. My biggest problem with the Farm was conceptual. My existing
wiki suddenly became a field inside the PmWiki base install. It seems to
me that much of this confussion goes away if PmWiki uses the farm
concepts by default. Then there's no terminology required. We just have
many wiki's using the same PmWiki code base.

We'd need to change the installation procedure, to include definition of
the location of the first wiki, but I don't think it even becomes more
confusing for first time installers.

  ~ ~ Dave


Joachim Durchholz wrote:
> Patrick R. Michaud schrieb:
>> [lots of dismissals as "this isn't a large problem"]
> 
> Now while I agree that none of these problems are showstoppers, each of 
> them can hinder an admin. It will take him an hour to find the cause, 
> research a solution, and implement it - if things go well; the process 
> can take days (if the symptoms don't directly point to the cause, or if 
> the PmWiki installation isn't publicly accessible and we can't look at 
> the ?action=diag output).
> 
> I'll grant that one can sensibly disagree about the severity of the problem.
> 
> However, I still have one question to ask:
> 
> What's the advantages of fields over installing multiple wikis with 
> shared code?
> (Or is it indeed just a terminological problem, and farms are already a 
> shared-code thing?)
> 
> To expand on the idea of "common code wikis", here's how I'd expect a 
> wiki software to work (and that's just my personal expectations, as a 
> datapoint, and - maybe! - something that PmWiki might approach in the 
> future, though of course there are other things than my personal 
> preferences to consider):
> 1) There's one directory for the "wiki engine" and one directory for the 
> "wiki data".
> 2) The data directory contains the index.php page that starts the 
> machinery. index.php calls the wiki engine with any parameters required 
> to make it find the data directory (ideally, the wiki engine would 
> simply look at the current directory).
> 3) The engine takes all data from the work directory, falling back to 
> the engine directory whenever the work directory comes up empty. (If 
> this is applied to scripts as well, this can serve to override bits and 
> pieces of the engine with wiki-specific code and configuration.)
> 
> I don't think it would be difficult to get PmWiki to such a scenario. On 
>   a safe_mode-disabled server, this would allow webmasters to install 
> the software, and nail down the configuration so that it "simply works" 
> for their customers.
> Being part of a web hoster, I can say that such a setup would make 
> PmWiki instantly attractive for us. (Not that it isn't already - but we 
> could then say things like "your WWW site comes with a preinstalled wiki 
> that you can clone if you need multiple wikis", instead of "install and 
> administer your own wikis". We could preinstall recipes that would make 
> sense for our customers. That all would definitely make a difference.)
> 
> Regards,
> Jo
> 
> _______________________________________________
> pmwiki-users mailing list
> pmwiki-users at pmichaud.com
> http://host.pmichaud.com/mailman/listinfo/pmwiki-users
> 




More information about the pmwiki-users mailing list