In the end of august the catalog to download packer binaries was broken because the recorded URL was something like
It is now again broken because it is
this change (# -> % -> : ->) is regularly done by Bintray to avoid web crawlers
After discussing with JFrog team the solution is to avoid to do HTML scraping and to use bintray APIs
3 tools installers catalogs are using bintray nowadays:
Only SBT script is using bintray APIs.
groovy and packer are using HTML scraping and should be updated to use Bintray APIs to be sure to not hardcode any URL pattern.
Note that it won't fully solve the problem of the inability to download packer (and groovy) installers because:
- Our catalog is updated every 4h thus we may have a delay where URLs in catalog are broken because they changed on bintray side
- AFAIK users don't have the ability to force the update of catalogs on a jenkins instance. It occurs when Jenkins is restarted (and maybe periodically ?). Thus they'll have to wait for a restart (or scheduled update) to retrieve the fixed version of a catalog