Bug#434358: cl-sql - arch-any without need
Package: cl-sql Version: 3.8.6.1-1 Severity: serious * debian/control: Change architecture of binary packages which depend on cl-sql-uffi from 'all' to 'any' since cl-sql-uffi is 'any'. Without this change, packages which depend on cl-sql-uffi can become uninstallable if the autobuilder fails to build the 'any' cl-sql-uffi and the dependant 'all' package gets upgraded. (closes: 433907) An arch-any dependency is no reason to make a package also arch-any. Packages are only moved into testing if they are up-to-date. Bastian -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#434358: cl-sql - arch-any without need
Bastian Blank wrote: An arch-any dependency is no reason to make a package also arch-any. There's a clear reason below. Whether you agree with the reason is a separate issue. Packages are only moved into testing if they are up-to-date. That's true. However, the change I made fixed a bug for a sid user where the all package was uninstallable because the dependant any package was not built by the autobuilder because of some lengthy delay. With the change I made, now sid will always have the correct packages at all times. Having that consistent working in sid is worth something. Is it worth the same increase in size of the archive (these packages are quite small)? For sid users (of which there are many), I'd say the answer is yes. -- Kevin Rosenberg [EMAIL PROTECTED] -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#434358: cl-sql - arch-any without need
Kevin Rosenberg wrote: Bastian Blank wrote: An arch-any dependency is no reason to make a package also arch-any. There's a clear reason below. Whether you agree with the reason is a separate issue. Packages are only moved into testing if they are up-to-date. That's true. However, the change I made fixed a bug for a sid user where the all package was uninstallable because the dependant any package was not built by the autobuilder because of some lengthy delay. Which is a perfectly normal situation for all the Arch: all packages that depend on Arch: any packages. With the change I made, now sid will always have the correct packages at all times. Having that consistent working in sid is worth something. Is it worth the same increase in size of the archive (these packages are quite small)? For sid users (of which there are many), I'd say the answer is yes. Not following policy because of some sid users' inconvenience is not the way to go IMHO. Note that we are thinking about fixing it globally somehow... Note also that you are trying to fix problems for sid users by introducing a workaround for all suites, so when you compare with the increase in the size of the archive you should count all suites. Cheers Luk -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#434358: cl-sql - arch-any without need
Luk Claes wrote: [...] Not following policy because of some sid users' inconvenience is not the way to go IMHO. Note that we are thinking about fixing it globally somehow... That makes sense and is good news. Thanks for the info. Kevin -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
