Hi David,
it seems that the version of cURL I was using did not have the requisite CA certificate, so it was not able to communicate via https.
I found the fix to my issue among several workarounds suggested on the following page:
- Brew Install results in Curl SSL certificate problem on Formula with GitHub HTTPS tarball URL
(see comment by "korny")
I believe this only happens on Mac machines using MAMP or "MAMP PRO".
The fix is remove the MAMP cURL library which is usually located at "/Applications/MAMP/Library/bin/curl". You can do this quickly, if you open a Mac terminal window and copy and paste the following command:
- mv /Applications/MAMP/Library/bin/curl /Applications/MAMP/Library/bin_curl
You may also have to restart MAMP (I did). After doing so, phpinfo reported the same version of cURL as I showed earlier.
In addition I had to reset the permissions on the appropriate folders:
- chmod -R 777 /path/to/moodle/mod/hotpot
- chmod 777 /path/to/moodle/mod
And "hey presto" it worked. David, congratualations on this excellent implementation of a most useful functionality.
I would like to humbly suggest that title of the mdeploy error page is changed to something that inspires confidence more than "Oops, it did it again" Also, I think admins would find a link from the error page to the mdeploy.log useful, or at least a copy of the last few lines of the log displayed right on the error page, so that they can begin diagnostic investigations more quickly. "Could not download zip file" was a mildly helpful message, but "cURL error 60 SSL certificate problem" enabled me to track down a solution right away. Finally, if my understanding is correct, then my original version of cURL could not ever communicate via https, so if it is possible to detect that situation automatically and report it nicely, that would the icing on the cake !
Joseph, while I was investigating this, I used the browser's back and forward arrows to skip between the steps in the deployment process, and I did see the "passphrase" error at one point. Were you doing something similar? When I did a "clean" run of the deployment, simply going from one step to the next by pressing the next/continue buttons, everything worked as expected.
best regards
Gordon
