OK. I deleted and restored. No visible change. I wonder if the issue is I have never noticed that the in-run version is lacking features? I've used the script in run only a couple of times so that's maybe 10 views in contrast to many, many more in aftercore.
I bristle when someone talks about a problem, claims it is frequent, claims it is mafia's problem and I have never heard of it. I accept that reaction is my problem but language is important.
That said I saved phulin-shorts-ui-branches-master-build before and after the delete. mafia said it was up to date before the delete so I expected no differences. I found a few files of the form bundle.number.* were missing and replaced(?) by similarly named files with a different number. That is a legitimate concern since mafia thought things were up to date.
In a pure svn environment I have seen this before in two cases. One is where svn doesn't really understand that something was renamed or that something was deleted. The cause was almost always operator error when checking in. (More later). The other case is when svn was being asked to manage an archive or a bundle that was subsequently unpacked. It was quite possible to change the contents in a way that svn did not detect as a file change and act accordingly.
The operator error may be something in the wrapper that maps between svn and git. Given the reluctant support for git in mafia it may also be that there is something that needs to be tweaked in SVNKit or maybe there is a update to SVNKit mafia needs to install. Perhaps, with mafia's new found willingness to use someone elses jar file we should investigate pulling SVNKit source out of lib?
In any event, I agree there is a problem, I am not sure I would describe it as a kolmafia caching problem, and I would first look at git->svn->mafia in hopes that it can be fixed by a change in process instead of code.
I bristle when someone talks about a problem, claims it is frequent, claims it is mafia's problem and I have never heard of it. I accept that reaction is my problem but language is important.
That said I saved phulin-shorts-ui-branches-master-build before and after the delete. mafia said it was up to date before the delete so I expected no differences. I found a few files of the form bundle.number.* were missing and replaced(?) by similarly named files with a different number. That is a legitimate concern since mafia thought things were up to date.
In a pure svn environment I have seen this before in two cases. One is where svn doesn't really understand that something was renamed or that something was deleted. The cause was almost always operator error when checking in. (More later). The other case is when svn was being asked to manage an archive or a bundle that was subsequently unpacked. It was quite possible to change the contents in a way that svn did not detect as a file change and act accordingly.
The operator error may be something in the wrapper that maps between svn and git. Given the reluctant support for git in mafia it may also be that there is something that needs to be tweaked in SVNKit or maybe there is a update to SVNKit mafia needs to install. Perhaps, with mafia's new found willingness to use someone elses jar file we should investigate pulling SVNKit source out of lib?
In any event, I agree there is a problem, I am not sure I would describe it as a kolmafia caching problem, and I would first look at git->svn->mafia in hopes that it can be fixed by a change in process instead of code.
