Bug#359644: adding a package should be atomic

2006-03-29 Thread Duck
Coin, "Bernhard R. Link" <[EMAIL PROTECTED]> writes: > This is quite difficult and the current emphasis is to try to never > delete anything that might cause high costs to retrieve (bandwidth) or > recreate (build). The problematic part is that makeing sure everything > will work before is only

Bug#359644: adding a package should be atomic

2006-03-29 Thread Bernhard R. Link
* Julien Valroff <[EMAIL PROTECTED]> [060328 15:57]: > >Are those not in the database already? Unless that got missed, > >a reprepro deleteunreferenced should remove all that stuff. > > Maybe this can be run automatically after such an error occurs (or best, > there could be an option in the the c

Bug#359644: adding a package should be atomic

2006-03-28 Thread Julien Valroff
Le mar, 28 mar 2006, "Bernhard R. Link" <[EMAIL PROTECTED]> évrivait : [...] (even if the db is clean, the filesystem should be clean too). Are those not in the database already? Unless that got missed, a reprepro deleteunreferenced should remove all that stuff. Maybe this can be run automati

Bug#359644: adding a package should be atomic

2006-03-28 Thread Bernhard R. Link
* Marc Dequènes <[EMAIL PROTECTED]> [060328 11:04]: > In the problem i encountered in #357803, adding a file built without -sa > for a non -1 package resulted in reprepro complaining for missing source > tarball, but with other files added into the pool. Either reprepro > should copy files after al

Bug#359644: adding a package should be atomic

2006-03-28 Thread Duck
Package: reprepro Version: 0.8.1-1 Coin, In the problem i encountered in #357803, adding a file built without -sa for a non -1 package resulted in reprepro complaining for missing source tarball, but with other files added into the pool. Either reprepro should copy files after all checks, or re