MarkE

You are almost certainly right. I needed a quick fix! I'm using POP3 at the moment. I'll try your suggestion when I have a bit more time. Ta.

The problem (or, at least, on of them) is line 505 of fMailbox.php. Since php 8.0.0, providing a non-string as a needle to strpos is no longer supported. $filter has a default value of null. (strpos($header, $filter) !== FALSE) evaluates to true and should be replaced. The following works in my case, but further fixes may be required if the $filter is actually set. if ($filter && strpos($header, $filter) !== FALSE)

I've now moved that site to php 8 and the module has stopped working. Any insights would be great. Meanwhile I will try and diagnose.

Simples: In init.php wire()->addHook('LazyCron::everyDay', null, 'batchRun'); and then in your hook: wire()->log->prune('debug', $days); where $days is set however you want In my case I aded a cronjob to access the site at night so that the LazyCron fires then.

That’s what I do. Part of a regular cron batch job.

That’s good (although @ryanobviously wanted to include it for some reason). The namespacing issue is still a problem, though, so I think the fix is needed.

UPDATE: Web service now seems to be starting OK. The problem might be to do with the docroot. config.yaml is inside site-admin/.ddev/ but index.php is of course in the root directory above site-admin. This is why I tried docroot: ../ (I have also tried ../../ as I'm not sure where the root is supposed to be relative to - .ddev or site-admin). However, if I go into ddev with ddev ssh and ls, it lists the members of site-admin, not the root. ddev logs shows an error of directory index of "/var/" is forbidden

I have a PW project with two sites - in site-admin and site-web respectively. I have moved the project from laragon to ddev, as I have done successfully with a number of (single-site) projects. However, this one won't run - I just get a nginx 403 error. I suspect it is something to do with the way I have set up ddev to deal with the multi-site. Because I wanted a separate url for site-admin and site-web, I did the ddev config within the site-admin directory, rather than the parent (root) directory. I then set the docroot to ../ so that it would pick up index.php. I have set index.config.php to point the url supplied by ddev to the site-admin directory. ddev start works OK, but I note that ddev describe does not show a web service even though it does show the urls. config.yaml is in site-admin/.ddev and contains name: BawdHall-admin type: php docroot: ../ php_version: "8.0" webserver_type: nginx-fpm router_http_port: "80" router_https_port: "443" xdebug_enabled: false additional_hostnames: [] additional_fqdns: [] database: type: mariadb version: "10.4" nfs_mount_enabled: false mutagen_enabled: false use_dns_when_possible: true composer_version: "2" web_environment: [] nodejs_version: "16" ddev describe shows ┌───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐ │ Project: BawdHall-admin ~/projects/BawdHall/site-admin https://bawdhall-admin.ddev.site │ │ Docker provider: docker 23.0.2 │ │ Router: traditional │ ├────────────┬──────┬──────────────────────────────────────────────────────────────────────────────────────────────┬────────────────────┤ │ SERVICE │ STAT │ URL/PORT │ INFO │ │ │ │ │ NodeJS:16 │ ├────────────┼──────┼──────────────────────────────────────────────────────────────────────────────────────────────┼────────────────────┤ │ db │ OK │ InDocker: db:3306 │ mariadb:10.4 │ │ │ │ Host: 127.0.0.1:32796 │ User/Pass: 'db/db' │ │ │ │ │ or 'root/root' │ ├────────────┼──────┼──────────────────────────────────────────────────────────────────────────────────────────────┼────────────────────┤ │ PHPMyAdmin │ OK │ https://bawdhall-admin.ddev.site:8037 │ │ │ │ │ InDocker: dba:80,80 │ │ │ │ │ `ddev launch -p` │ │ ├────────────┼──────┼──────────────────────────────────────────────────────────────────────────────────────────────┼────────────────────┤ │ Mailhog │ │ MailHog: https://bawdhall-admin.ddev.site:8026 │ │ │ │ │ `ddev launch -m` │ │ ├────────────┼──────┼──────────────────────────────────────────────────────────────────────────────────────────────┼────────────────────┤ │ All URLs │ │ https://bawdhall-admin.ddev.site, https://127.0.0.1:32797, http://bawdhall-admin.ddev.site, │ │ │ │ │ http://127.0.0.1:32798 │ │ Any ideas?

Added a fix for the on-demand mirroring at the beginning of the hook: //Don't execute for backups!! $backtrace = DEBUG::backtrace(); foreach($backtrace as $item) { if(strpos($item['file'], 'ProcessDatabaseBackups')) return; }

Just posting this here as I raised an issue some time ago about TextformatterHannaCode module, which has not been responded to or addressed (see https://processwire.com/talk/topic/3745-hanna-code/page/18/?tab=comments#comment-217245 and https://github.com/ryancramerdesign/ProcessHannaCode/issues/26). I have also recently raised a PR (https://github.com/ryancramerdesign/ProcessHannaCode/pull/27) to fix it. Essentially the problem is that, if you have a PHP Hanna Code which is namespaced (ProcessWire), then the module will remove the namespace (irritating in itself), but also (and maybe more seriously) it will omit the line if(!defined("PROCESSWIRE")) die("no direct access"); which is potentially a security weakness. The PR is a simple one-word fix.

A slightly odd case this: I have a WSL/Docker/ddev setup that has been working very well. Then, for no apparent reason, Docker Desktop failed to start the docker engine. I tried all the suggested fixes without success so decided to uninstall it and run Docker from within WSL instead. After successfully doing that, I found that the db was empty so PW would not run. I imported a recent backup database from the production site into the empty db using phpMyAdmin and it appeared to be OK. However, on running the PW site, I got a 500 error. I tried importing an old dev backup and that worked OK. So, now I had PW running in dev, but not with the latest database. I then tried to use the PW backup tools to download the latest version from the live site and upload it into the dev site and got the following message after clicking 'upload': ProcessDatabaseBackups: File could not be downloaded (https://[Database path].sql) 403 Forbidden: (tried: curl) Any idea what might be causing this and how to fix it? At the moment I am stuck without being able to work on a current database version and can't get it successfully imported via either phpMyAdmin or PW. One thing that looks rather odd is that the message says it can't be downloaded (not uploaded) and that the database path given is a path to site/assets/files on the LIVE site. EDIT: Also, I see that the uploaded backup IS listed on the backups page. However, restoring it results in errors*. All a bit random and confusing! * The error appear to be because not all modules are loaded. Commenting out the calls and refreshing the modules seems to fix this.

@ryan - this still seems to be open. I assume the intention is not to add the namespace if it isn't in the hanna PHP code, otherwise it should be added and so should the 'die'. I think, in that case, the fix is simple - replace the line `$code = str_replace($openPHP, "$openPHP\n$firstLine\n", $code);` by `$code = str_replace($php, "$openPHP\n$firstLine\n", $code);` The existing line does not work because, if namespace was in the code it has been removed, but $opnPHP includes it, so the str_replace does not operate. I have created a PR for this ----------------- If you do not have namespace declarations in your hanna code php and need them, then the following executed in Tracy console seems to do the trick: $h = $modules->get('TextformatterHannaCode'); $codes = $h->hannaCodes(); $codesAll = $codes->getAll(); foreach($codesAll as $code) { $codeString = $code->code; if(!strpos($codeString, 'namespace')) { $codeString = str_replace('?php', '?php namespace ProcessWire;', $codeString); } $code->code = $codeString; $codes->save($code); }

Does anybody use this and/or have a guide for using it with ProcessWire? There are guides for various CMSs and frameworks and they are open to more.

Interesting question. I am increasingly putting hooks in module ready() methods rather than ready.php, which naturally leans to the second approach, so I’d be interested if anyone has benchmarked this.

Another thought, you may want to record invoices separately from orders, to deal with unsatisfied or partially satisfied orders. Alternatively this might be handled by status fields on order lines. At the risk of repeating myself, getting the business specification right is paramount. PS I absolutely agree with @teppo regarding decimal fields. A slip on my part.

Another reason for doing some deep thinking about the entity model before choosing. If payments can only ever belong to an order (and only one order) then they should be children of the order otherwise they risk being orphans. If payments can cover more than one order, then they should stand alone, but you need to do serious thinking about how they are allocated to each order if the amounts don’t match exactly. This might be achieved via an “allocation” field (float or decimal) in the order and the payment. And/or you might need to split orders down into order lines. An accurate understanding of the business model is required. Plus how it might evolve. Then do the entity modelling. Lastly decide how to implement in PW. And don’t forget that @Robin S‘s excellent ConnectFields module is just the biz for maintaining many-many relationships (e.g. between the two allocation fields referred to above).

I would think it all depends on the cardinality of the relationships. If one-many then children are fine. Otherwise group separately and probably use ConnectFields module.

Not sure why @ryanintroduced that. Should we request reversion?

Nothing really. You just make sure PhpStorm is listening then run “ddev debug”.

Probably worth mentioning that since moving to ddev, I have found that xdebug works well with PhpStorm too. Much better than Laragon. My mileage!

Something like this hook in your ready.php file? wire()->addHookAfter('InputfieldSelect::render', function($event) { $inputfield = $event->object; if($inputfield->name == 'my_select_field') { if(count($inputfield->options) == 1) { $inputfield->description = "text for one item"; $event->object = $inputfield; $event->return = "<p>{$inputfield->options[1]}</p>"; } else if(count($inputfield->options) == 0) { $event->return = "<p>No options!</p>"; } } }); EDIT: Added condition so that it only operates on the relevant inputfield. Also, note that while this fixes the UI, you will need to make sure the value is provided where needed as it is no longer selected (may be possible to address this by making the value required and providing the default value). EDIT2: For example, you could set the value like this: $page->of(false); $inputfield = $fields->get('my_select_field')->getInputfield($page); if(count($inputfield->options) == 1) { $page->test_select = $inputfield->options[1]; $page->save('my_select_field'); } Is that any help @The G?

I now have my project files in the WSL2 environment and that does seem to improve things without mutagen. I also tried running PhpStorm in WSL2 but it was a bit of a pain and no quicker.

Thanks for all the ideas @Jan Romero . Personally I got a bit frustrated with TrackChanges because it seems to be a bit random about what it records - for pages as well as fields etc. That’s why I started using getFresh() for pages with a home-built compare function because it gave me full control and understanding. So my idea was to do something similar for fields and templates (and fieldgroups) - the code above is my attempt at this and no doubt can be improved. The getFresh() for pages is hugely more complex! Nevertheless, I agree that a more robust and consistent TrackChanges is desirable.

That seems to work and might be sufficient if you just want to know if there are any changes. The test that I did - adding a field to a template - gave a curious array of changes in the bar dump, viz: array 'compile' => array 0 => 3 'noParents' => array 0 => '' i.e. nothing about the field that had been added. I have no idea why this is. On the other hand, my getFreshSaveableItem() method returns the complete template before any changes, so includes the related fieldgroup. However it may be slower - I haven't benchmarked them.

I stopped doing that a while back because it was taking longer to recognise code changes. But it may have been down to some problems I was having with PHPStorm, so I’ll try again now those problems seem to be fixed ?. Thanks for the nudge.