in #32 autojobs was extended by the possibilty to use wildcards (*) in the branches configuration.
We introduced an additional folder level for better grouping of the projects in svn. Then we had a similar problem as in #32 because we don't want to configure one branches url for each category and add a new entry whenever a new category is created. So we just starred (*) this folder depth as well.
Without the additional folder our branches url's looked like this:
http://host/path/to/REPO/java/*
http://host/path/to/REPO/java/*/branches
The folders at the * level were our projects, each containing a svn standard directory layout (trunk, tags, branches).
With the new additional folder level for categories, we tried to use those url's
http://host/path/to/REPO/java/*/*
http://host/path/to/REPO/java/*/*/branches
Although in the implementation of #32 the wildcard lookup is called recursively, the handling of the method result is not considering this use case.
For this project
http://host/path/to/REPO/java/category1/projectA/trunk
using
- http://host/path/to/REPO/java/*/*
results in an error when autojobs tries to process this "detected" branch:
java/category1/*/java/category1/projectA/trunk
After analysing the problem it looks as if it could be easily fixed by just adding a case distinction after the recursive call of svn_wildcard_ls.
With this patch there is no problem with the configuratin from the example above and for each project in each category all trunk and branch jobs are created correct.
It would be great if you could apply this patch into master for the next release.
Hi,
in #32 autojobs was extended by the possibilty to use wildcards (*) in the branches configuration.
We introduced an additional folder level for better grouping of the projects in svn. Then we had a similar problem as in #32 because we don't want to configure one branches url for each category and add a new entry whenever a new category is created. So we just starred (*) this folder depth as well.
Without the additional folder our branches url's looked like this:
http://host/path/to/REPO/java/*
http://host/path/to/REPO/java/*/branches
The folders at the * level were our projects, each containing a svn standard directory layout (trunk, tags, branches).With the new additional folder level for categories, we tried to use those url's
http://host/path/to/REPO/java/*/*
http://host/path/to/REPO/java/*/*/branches
Although in the implementation of #32 the wildcard lookup is called recursively, the handling of the method result is not considering this use case.For this project
http://host/path/to/REPO/java/category1/projectA/trunk
using- http://host/path/to/REPO/java/*/*
results in an error when autojobs tries to process this "detected" branch:java/category1/*/java/category1/projectA/trunk
After analysing the problem it looks as if it could be easily fixed by just adding a case distinction after the recursive call of svn_wildcard_ls.
Please find the bugfix attached to this issue. patch_multiple_wildcards.zip
With this patch there is no problem with the configuratin from the example above and for each project in each category all trunk and branch jobs are created correct.
It would be great if you could apply this patch into master for the next release.
Best regards Alex