C Coffey

Concur: And just so you don’t think we’re debating, I think the combination of my questions and your answers is probably going to prove very useful to people who have struggled with this. We have to remember, a good part of the community are not technical, and probably don’t want to be.

Concur on separation of code, data, and configuration. Point remains pulling the prd DB back to dev gives a very representative set of content to experiment with. For some elements of look and feel you need sufficient quantities of articles, categories, pages, etc, to get it right (right being how you want it).

Of course, instead of users having to backup their content folders, MySQL DB, and do migration hacking, it would be really useful if WordPress had a ‘backup up the site’ option which captured the MySQL, content, themes and WordPress code all in one backup. I see the feature isn’t native, but there are plugins to take care of much of this, so, no major problem. Not sure if those plugins take care of the URL customisation requirements, will find out soon enough though.

Another unknown for me:

If I install a theme, and customise it, and then install another theme, and customise it, are the customisations I make to both themes all saved inside the DB, ie: I can flip from one to the other?

When I say ‘customise’, I mean, using the settings provided by the theme inside the dashboard, not hacking of theme PHP/HTML/CSS do extend it.

Can I experient with a new theme ‘offline’ some how and then activate it?

Possible to move just the theme settings from ‘dev’ to ‘prd’ so that I can reskin by live site after testing out the theme in development?

These are things I can suck-and-see in the next few days, though it will probably be helpful to other people to have the answers here in the forum.

Seems some of the issues here are:

– Word press has more than one method of storing and managing content (ie: mysql and flat file system, I DO NOT like that …).
– Word press allows configuration of the domain/path but this has of necessity to exist in multiple places.
– Absolutely links are preferable in certain situations (see useful analysis here https://www.ads-software.com/support/topic/using-relative-instead-of-absolute-links-for-images).
– But, in my view an absolute link should be presented outwardly by the site [generated] with reference to a single point of reference for what the site root domain and path are.
– The domain and path should not (IMO) be embedded throughout the content of the site in the tool itself. To the outside world, they should of course appear consistently to be fully qualified.

I have seen several hacks for fixing up a site intended for one domain/subdomain/path to work in another. Assumably that involves updating base config. data in WP, both in the FS and DB, but also searching and replacing for links in both the FS and DB (and including re-tailoring WP files, eg: wp-config.php). This is hardly elegant.

Given how things stand today, I’m wondering is the best solution for this to actually make a test site ‘believe’ it is the production site, ie:
– Create a batch file/script that adds/removes an entry in (for example the windows case) for mydomain in \windows\system32\drivers\etc that actually resolves “mydomain” back to “localhost” so that the site runs locally as if it were at “mydomain”.
– Taking that one step further, create an isolated VM sandbox (Linux) with Apache/MySQL/PHP, etc, where the sandbox is isolated and believes it is “mydomain” but where the test site files are actually mounted in from the host to a location (as appropriate) in the guests file system.

I’m on day 2 of this exercise, so it’s unclear to me (for example) if the ability to override default DNS routing other than at OS level is a feature of web servers in general, to support testing, but I don’t think it’s a long stretch that this is true. So, even without hacking routing, or creating a deluded VM, is it possible to configure Apache, or IIS express, or Expression Web Development Server, to subvert a DNS request and route it back to localhost?

Some good points were made in other posts – you never want to promote test ‘data’ to production – but – production data is a very good source of test data [subject to business need to know/confidentiality agreements, etc].

In this case, it should also be possible to subvert the domain for the WP production site, but not for the production mySQL database. Now, that would be suicide for a corporate application, but not for someone developing a blogging web site. So, in theory it should be possible to subvert WP, but not mysql, assuming they are not sharing the same domain name.

What I need to check, is, does subversion of a domain also imply automatic subversion of subdomains, ie:
mydomain.com
sql.mydomain.com

But, it given helm and other VWS it’s not a long shot to use different primary domains for WP and DB, ie:

mydomain.com
sql.myotherdomain.com

Finally, whilst we could point a ‘dev’ copy of the FS artifacts at a ‘prd’ copy of the DB, different WP themes suggest a need for different internal structures. So, it is probably not always feasible to test ‘dev’ themes against ‘prd’ data. The other issue I need to address here is a push button method of refeshing a ‘dev’/’tst’ DB from the ‘prd’ DB.

Will update if I find anything useful on the above points.