adrian

8 hours ago, Mike Rockett said:

@adrian – I've released 1.5.64 which adds the safety net around CLI usage (as a general rule) and request_uri. Please let me know if that solves it.

Thanks - that has taken care of things. I actually think there are lots of modules that get loaded via CLI scripts even though they aren't needed. I kinda think PW should have a module autoload option for this to make it easier / more obvious to the developer to disable via CLI unless needed.

40 minutes ago, Mike Rockett said:

The only thing I can think of is an upgrade to PHP 8.1, as that's the only thing that can cause it.

That would make sense, except I have been running 8.2 on that server for many months and the error started appearing immediately after updating Jumplinks from 1.5.61 to 1.5.63.

35 minutes ago, Mike Rockett said:

Guessing this should do the trick…?

Yep, that should work just fine.

Just a little FYI about something I have noticed in the past - sometimes $_SERVER['REQUEST_URI'] returns false rather than the URL or NULL (https://github.com/nette/tracy/issues/351) which means checks need to consider that. Obviously not relevant given your CLI check.

PS - thanks so much for continuing to support your modules even though you're no longer using PW - very generous indeed.

@Mike Rockett - sorry, I should have come with more details - looks like it's happening from a cron executed PHP script that bootstraps PW. That would explain why it's null.

I'm not sure why it's a new error though - on a quick scan I can't see anything from your most recent commit that would cause this change.

Hi @Mike Rockett - brand new issue with the latest version: 

PHP Deprecated: ltrim(): Passing null to parameter #1 ($string) of type string is deprecated in ProcessJumplinks.module.php:206

EDIT: I see that line hasn't changed recently, but I've never seen the error before on the same server with no recent PHP upgrades. Regardless, it still needs a fix please and thanks.

Have you tried a hard reload in the browser? I know it sounds strange, but I have seen PW return cached results before. Please let me know if that works.

Actually, something to consider - what about on Windows - wouldn't that need the extra slash because instead of the filepath starting with a slash it would start with a drive letter.

14 hours ago, bernhard said:

PS: Just did an update to the latest version and got this error:

Error: Undefined constant Tracy\Debugger::Version in /var/www/html/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Bar/assets/bar.phtml:15 Stack trace: #0 /var/www/html/site/assets/cache/FileCompiler/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Bar/Bar.php(119): require() #1 /var/www/html/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Helpers.php(364): Tracy\Bar->Tracy\{closure}() #2 /var/www/html/site/assets/cache/FileCompiler/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Bar/Bar.php(120): Tracy\Helpers::capture(Object(Closure)) #3 /var/www/html/site/assets/cache/FileCompiler/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Bar/Bar.php(89): Tracy\Bar->renderPartial('main') #4 /var/www/html/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Debugger/DevelopmentStrategy.php(123): Tracy\Bar->render(Object(Tracy\DeferredContent)) #5 /var/www/html/site/assets/cache/FileCompiler/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Debugger/Debugger.php(309): Tracy\DevelopmentStrategy->renderBar() #6 [internal function]: Tracy\Debugger::shutdownHandler() #7 {main} (stored in /var/www/html/site/assets/logs/tracy/exception--2023-11-05--11-52--6565da4fb6.html)

Yeah, I know about this and not sure there is an easy way around it, but it fixes itself after a modules refresh.

Hey @bernhard - you can change this in the module settings in the "Editor Links" section.

Just change: 

vscode://file/%file:%line

to:

vscode://file%file:%line

I'll change the default value and note in the next version.

At a quick glance, it feels like you are missing a level when getting the video. You need to get the video_field from the select_video, but instead you are getting video_field from $item. Does that sound right?

9 hours ago, flydev said:

I was also thinking of pushing a new feature that would allow you to export and import module settings on other installations, what do you think, it could help? 

I always have https://processwire.com/modules/module-settings-import-export/ installed on all my sites, so no need from my end for it to be built into your module.

Actually, the only other suggestion is to make it a real initial value for the textarea rather than a placeholder, because I actually had to go and open mysqldump.unix.sh to get a starting script I could then edit.

Thanks @flydev 

The only issue I noticed is that the supplied mysqldump.unix.sh looks like this:

#!/bin/sh

# (1) Set up all the mysqldump variables
FILE=%%FILE%%
DBSERVER=%%SERVER%%
DATABASE=%%DATABASE%%
USER=%%USER%%
PASS=%%PASS%%
CACHEPATH=%%CACHEPATH%%

# Fix trailing slash in cache path
CACHEPATH="${CACHEPATH%/}/"

OUTPUT="${CACHEPATH}${FILE}"

# (2) In case you run this more than once a day, remove the previous version of the file
# unalias rm     2> /dev/null
# rm ${FILE}     2> /dev/null
# rm ${FILE}.zip  2> /dev/null

# (3) Do the mysql database backup (dump)
#  - to log errors of the dump process to a file, add --log-error=mysqldump_dup_error.log
# (a) Use this command for a remote database. Add other options if need be.
# mysqldump --opt --protocol=TCP --user=${USER} --password=${PASS} --host=${DBSERVER} --port=${PORT} ${DATABASE} ${DATABASE} > ${OUTPUT}
# (b) Use this command for a database server on localhost. Add other options if need be.
mysqldump --routines --triggers --single-transaction --user=${USER} --password=${PASS} --databases ${DATABASE} > ${OUTPUT}

Note the missing PORT from the variables in #1 and also the duplication of ${DATABASE} in #3a

Once I fixed those two things, it work great!

Thanks again for making this configurable - a really nice improvement!

Hey @Robin S - good question. I have played around a little without any success. I feel like I am missing something obvious like output buffering getting involved, but perhaps it is just down to the way I instantiate a PW template and render it via AJAX https://github.com/adrianbj/TracyDebugger/blob/ec79b76bf0443118a6c402c9ffe2a9c491a8b5cc/includes/CodeProcessor.php#L197-L203 

I'll keep mulling it over, but not sure what the solution might be at the moment.

Thanks - looks like it goes deeper than php-saml to one of it's dependencies (https://github.com/robrichards/xmlseclibs) - https://github.com/SAML-Toolkits/php-saml/issues/79

php-saml branch without mcrypt: https://github.com/SAML-Toolkits/php-saml/tree/remove_mcrypt but this was created in 2017, so who knows when it will make it into the master branch.

And it looks like mcrypt has been removed from https://github.com/robrichards/xmlseclibs - https://github.com/robrichards/xmlseclibs/pull/101

I don't know - looks like some of these relied upon packages aren't really being maintained anymore, so hopefully php-saml will merge those changes to master sometime :)

@BitPoet - thanks so much for your hard work on this and sharing it with everyone.

Just curious though about the need for the PHP-mcrypt library. I thought it was considered important to instead use sodium these days. I'd love your thoughts on this because it seems like you are pretty in-tune with these types of things.

Thanks @ryan - this is fantastic update. My initial testing shows that OR selectors are now working for the WireCache $expire selector (https://github.com/processwire/processwire-issues/issues/1828) which really opens up the ability for some huge performance benefits.

Thanks again!

Yeah, PW page objects are pretty huge. I wonder if you might also benefit from removing the Full Object tab from the dump and just going with the Debug Info tab? Personally I wouldn't because there are times I do want to look through the full object, but perhaps you might find it a better option?

That is the reason that d() is the standard method for dumping rather than db (dumpBig). d() uses those defaults, but db() overrides them.

Tracy module settings > Code Editor Settings:

Hi @flydev - for remote DBs, I find I need this for native backups to work.

    $data = '
        # (1) set up all the mysqldump variables
        FILE=' . $this->options['backup']['filename'] . '
        DBSERVER=' . wire('config')->dbHost . '
        DATABASE=' . wire('config')->dbName . '
        USER=' . wire('config')->dbUser . '
        PASS=' . wire('config')->dbPass . '
        PORT=' . wire('config')->dbPort . '

        # (2) in case you run this more than once a day, remove the previous version of the file
        unalias rm     2> /dev/null
        rm ${FILE}     2> /dev/null
        rm ${FILE}.zip  2> /dev/null

        # (3) do the mysql database backup (dump)
        mysqldump --opt --protocol=TCP --user=${USER} --password=${PASS} --host=${DBSERVER} --port=${PORT} ${DATABASE} > ' . $cachePath . '${FILE}
        ';

The key things are setting the DBSERVER to $config->dbHost and adding the PORT

Also, wondering if you've had any thoughts on making the line in section #3 configurable in the module settings, because currently any module updates break native backups.

Thanks!

@da² - unless I am misunderstanding you, the PW Upgrades module doesn't do anything to upgrade the DB. There are almost never changes to the PW DB schema, but if they are required, then it is handled automatically the first time PW loads after the contents of the wire folder are updated with the new version.

Hi @Paul Greinke - as @teppo said, it still works as intended on the most recent PW versions and I use it in production so I should be aware of any issues that might arise in the future.

19 hours ago, Robin S said:

This didn't work for me. $_SERVER['HTTP_HOST'] and location.hostname just give me "mywebsite.com" and without the "https://" it doesn't form a valid base for the URL.

That makes total sense, but for some reason it worked fine for me - weird.

Unfortunately $_SERVER['HTTP_ORIGIN'] doesn't work when the page 404s.

I did some testing with $_SERVER['PHP_SELF'] and that actually seems to work nicely, including when bootstrapping PW in a directly called .php file in the URL. However, reading back through https://github.com/nette/tracy/issues/407 I see that there is a specific reason for loading from the current URL (the linked to XDebug issue: https://github.com/nette/tracy/issues/216), so I am not sure this is actually a good idea.

Maybe it does come back to the last point I made in that issue and the loss of the GET params making it through. I am not sure anymore, and am out of steam on this at the moment. Let me know if you have any good ideas :)

Hey @Robin S - this has bugged me for quite some and I did ask about it some time ago: https://github.com/nette/tracy/issues/407

It's basically due to htaccess's rules 16 & 17 in combination with the way the Tracy core loads itself at the base url rather than from the root url, eg, it loads:

http://mywebsite.com/doesn't-exist/?_tracy_bar=js&v=2.10.3&XDEBUG_SESSION_STOP=1

rather than:

http://mywebsite.com/?_tracy_bar=js&v=2.10.3&XDEBUG_SESSION_STOP=1

I have been playing around with fixing that. It seems like everything is working, but it's not well tested yet, so I am not sure if there will be any side effects. Would you mind also testing?

/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Bar/assets/loader.phtml 

//$baseUrl = $_SERVER['REQUEST_URI'] ?? '';
$baseUrl = $_SERVER['HTTP_HOST'] ?? '';

/site/modules/TracyDebugger/tracy-2.10.x/src/Tracy/Bar/assets/bar.js

    //baseUrl = location.href.split('#')[0];
    baseUrl = location.hostname;

Just now, Tyssen said:

Not necessarily. Page title might be terms and conditions, but you don't really want the page name being /terms-and-conditions/

Agreed, that is why I said in almost all cases :)

That is also why the module has lots of options for respecting existing differences and excluding certain pages / templates from its effects etc. It's definitely worth playing with :)