| | 41 | |
| | 42 | Answer: |
| | 43 | |
| | 44 | In fact, there is no way to force the creation of subdirectories at "cmt create ..." time. |
| | 45 | |
| | 46 | The "branches" statement - as it is today - is useful in the following use case: |
| | 47 | |
| | 48 | o A subdirectory is only needed either at run time or at build time (for instance it will receive temporary files) |
| | 49 | o This subdirectory is not part of the sources themselves (and it's NOT in CVS) |
| | 50 | o Thus it must be "recreated" when the sources are eg checked out from CVS |
| | 51 | |
| | 52 | |
| | 53 | This indeed does not cover your use case which is to systematically create new packages with a given predefined structure. |
| | 54 | |
| | 55 | Probably one good example is when a convention exists to provide include files in the subdirectory |
| | 56 | |
| | 57 | ../<package> |
| | 58 | |
| | 59 | This would be useful to enforce this convention at "cmt create ..." time. |
| | 60 | |
| | 61 | I think your suggestion to describe this convention in the project file is a good idea. One implementation could be to accept the "branches" statement at the project file level, and give it the semantics you suggest. |
| | 62 | |
| | 63 | Let me add this to some next release of CMT. |
| | 64 | |
| | 65 | Thanks anyway for this point and the suggestion. |
| | 66 | |
| | 67 | Regards, |
| | 68 | Christian |
| | 69 | |
