On 11/18/2009 12:51 PM, Adam Williamson wrote: > I don't understand why everyone's tip-toeing around Firefox's > brokenness, here. Other applications don't seem to have problems with > in-place updates. Why isn't anyone filing bugs on Firefox upstream? Aren't we deleting the old firefox on upgrade? /usr/lib64/firefox-3.5.5/ is what I have now. /usr/lib64/firefox-3.5.4/ is gone. Is it reasonable to delete an application and expect it to continue to run properly? I think all Firefox could do to prevent this would be to open() all the files it might need so that when the filesystem marks them as deleted they'll still be usable until closed. But that means >450 open filehandles with the current version. Maybe the problem is that we're deleting the files out from under a running application. Or our directory naming conventions are wrong. Or our filesystems lack semantics to handle this elegantly. Maybe updates should be more event driven. The general class of problem for the app seems reasonable and creating an O(n) solution instead of an O(1) solution seems like the wrong approach. I suspect a bug filed on BMO "continue to work properly after Firefox is deleted" will get WONTFIX'ed or NOTABUG'ed pretty quickly. -Bill -- Bill McGonigle, Owner BFC Computing, LLC http://bfccomputing.com/ Telephone: +1.603.448.4440 Email, IM, VOIP: bill@xxxxxxxxxxxxxxxx VCard: http://bfccomputing.com/vcard/bill.vcf Social networks: bill_mcgonigle/bill.mcgonigle -- fedora-test-list mailing list fedora-test-list@xxxxxxxxxx To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-test-list
