Pierre-Luc

All the WP sites I did in the past (way too many of them, now I switched entirely to PW, I even redid whole clients simple "brochure" websites pro-bono to save me the hassle) have been ok, but there's always something with the API that doesn't work quite right (future dates in queries, good luck!), maintenance issues, compatibility issues (loading 10 versions of jquery and the like). Some things might work right when developping, but often I need to change the plugins code to improve them, which is a problem. Then when new versions of everything comes out, it's a mess. WP sites don't seem to grow very well either, so as complexity increases, it all falls apart. Speed issues are also a big thing and really hard to optimize to get good load times. All that time waiting for the admin to do stuff is wasted time in billable hours (I charge a minimum of 30 mins for maintenance, most things take less than 15 mins to do, the process is usually much faster in PW I've found so far). Also take into account that when you do full-custom work like I do (I'm a designer AND a programmer), WP always comes with either too many or not enough features. That's time I need to just do configuration to get rid of the unneeded parts, which sometimes isn't really straightforward. I also do a lot of multilingual things, WP can do it with plugins, but the workflow is kinda weird and involves a lot of support calls from clients. Some of the features don't work so well either. So far what I've launched with PW is trouble free. It does everything much better and faster. I can say with big relief that WP's custom post types are a thing of the past. I was able to implement different relatively "complex" things (well, would be in WP) like 100% custom event systems, payment/donation systems, etc (hopefully will be able to package some of them nicely to reuse later and share in modules) in next to no time. Now I'm happier, clients are happier, work is faster, everybody wins.

I've been dealing with an issue lately. Someone seems to have been querying one of my websites a *LOT* and been trying to run queries which give me admin notifications from PW. I get the typical error with page requested, username and error number and description, but I would like this error message to display the IP and possibly the query variables so I don't have to dig into logs and see where it comes from and possible breaches tried. Anyone knows if there's an easy way to add that info?

This might be a dead topic, but I'm looking at doing an implementation using Google Analytics (GA) API to do that on a website that has caching (and don't want to create too much custom cache code that I'll have to maintain afterwards), using a cron job. I looked at the documentation, and tried a few things with GA's Query Explorer : http://ga-dev-tools.appspot.com/explorer/ https://www.googleapis.com/analytics/v3/data/ga?ids=ga:YOUR_GA_ID // Set your request metrics here &metrics=ga:visits,ga:pageviews // Set a filter for your page, you can combine multiple paths into one request using commas (OR operator) and multiple pagePath values &filters=ga:pagePath=~/event/x/,ga:pagePath=~/en/events/x/ // start-date cannot go before 2005-01-01 (GA launch date) and end-date should be set to today &start-date=2005-01-01&end-date=2014-04-15 This will return json data that you can then insert into a field. All in all, it seems easy: Setup a script that will parse all pages, get multiple values, create a GA Api query, get values, update database fields. Note that there are limits and quotas to the API. For what I need it, it's not a problem, but you might need to reach support to get more info. Hope this helps some of you.

I recently found while doing work on Laravel a nice tool called Var Dumpling that will convert all var_dumps to nicely formatted blocks of text with a structure, making it a lot easier do debug var_dumps. It's available on Chrome and Firerox, you can see what it looks like on their page http://vardumpling.com/ Hope some of you find it interesting!

MariaDB pretty much has the same features MySQL has, and still has a problem that all calculated on a plane, which can lead to problems when calculating distances, mostly if points are very far from each other.. PostGIS is much more advanced, but I'm still considering if MySQL/MariaDB would do the job.

Thanks for your answer. There are big limits in regards to geodatabases in MySQL that I would prefer not to deal with. I'll see if there are ways to implement what I have in mind using MySQL without killing performance and making sure the calculations are ok. I hear there have been improvements in 5.5 but my go-to is usually PgSQL for that kind of work. Do you have in mind any requests or methods that might be very specific to MySQL in Wire?

Thanks Adrian, I added a comment on the issue.

Fixed an issue with embedding in secure pages. YouTube's oembed service doesn't return https urls, even if the feed url is https. Change line 86 to: $embedCode = str_replace('http','https',$data['html']);

I would think using repeaters on a template would establish relationships between "classes" (I feel strongly like templates behave like classes). Just add your tags on images, and search using the image title/name, and add criterias for tags.

Guest is a mandatory role for every users. Excluding guest will exclude all users. TLDR, never use !=guest.

Hi, I was wondering if any of you ever tried to run ProcessWire on PostgreSQL. Since 2.4, Ryan seems to have moved PW to PDO completely, and that would make things easier to change the database driver. From running quick greps on wire/ and site/ on a few installed projects, it seems like not too much hassle. I'm asking this because I'm thinking of implementing a few geometry fieldtypes and "geoselectors" using PostgreSQL/PostGIS. @Ryan, any thoughts on doing this?

I confirm, trying with the Selector test modules, this is working : template=user, roles!=affiliate, roles=editor Commas are the AND selector.

You're right, good points!

Hey David, really like AIOM. I would usually use Assetic for that, and I think you might want to look at integrating that code. Could save you a few headaches.

After a bit of testing it seems that it takes into account all the values in the repeater (which is ok for my use). But I'd still like to know if it's possible to use selectors that way.

Is there something like that available in PW ? I need to filter something in a repeater by its last value. Something like that: repeater.last.field repeater.first.field

Thanks a lot, hoping we'll see a 2.4.1 soon with a few fixes.

Is that so that the repeater doesn't support field dependencies Ryan ? This is mentioned nowhere in the documentation or in the UI. That would be extra helpful to have it mentioned on both the repeater and inputfield-dependencies pages if that's the case. What I meant to reproduce is the two time fields not saving any data whatsoever when there are visibility rules. I don't have anything weird installed : Images with cropping, Map Marker, CKE, HTML Purifier, Video embed...

I was able to isolate the bug and replicate it. I deleted all my fields including the repeater and started anew. The structure is as such: event_dates -- repeater, mandatory event_day -- date (Y-m-d) event_allday -- checkbox event_day_start_time -- inputtext with regex event_day_end_time -- inputtext with regex They all display at 50% to make a 2x2 grid. The bug happens when I put a visibility constraint on event_day_start_time and event_day_end_time of event_allday=0. This happens whether or not there is a display rule. I am unsure this bug also applies to other field types or when fields are outside a repeater where the visibility is also constrained by a checkbox or another kind of field. I'm in rush mode to have it go live by yesterday.

Nope, removing the field length, striptags and regex changed nothing.

No, they don't.

I have 2 simple text fields next to a checkbox and date picker, all into a repeater, which are not saving. They both represent a time value and have the following options activated : Striptags: On Size and MaxLength: 10 Pattern: ^(0?[0-9]|1[0-9]|2[0-3])(:[0-5][0-9])$ I've ran out of ideas on what could cause it. Any help is appreciated, I can give any of you access to the dev site if needed. This is on 2.4 stable.

Thanks Ryan, I saw the release just yesterday and updated my work directory. Great job!

Thanks a lot for your quick reply!

Hi, I've selected ProcessWire to implement a website. As it stands for now, 2.3 would do the trick but the improvements in dev would simplify a few of the implementation that are required. I was wondering if it was a good idea to use dev instead of 2.3 What kinds of stability can I expect from it and will there be a lot of breaking involved in the transition from 2.3 to 2.4? Thanks!