Hi, >imapcopy (1.04-2.1) unstable; urgency=medium
> * Non-maintainer upload > [Peter Michael Green] > * Fix build with fpc 3.0.0 (Closes: 818448) > > -- Gianfranco Costamagna<locutusofb...@debian.org> <date of upload> and this is something I do almost everytime, except when the work is already complete (I mean the patch is already a full debdiff) >This way it correctly identifies both who authored the change and who decided it was ready to upload to Unstable. just to be clear (even if I don't think there has been misunderstanding here) There are *many* ways to identify the uploader of a particular package (my favorite is downloading the changes and gpg it to see the signature) My intention was to avoid having to tweak changelog (laziness) and give you full credits, even if the testing has been performed by two other people >You should only upload with someone else's name in the changelog trailer if >they have explicitly asked for sponsorship. this makes sense, indeed >Maybe i'm overcautious but I will only NMU if I have tested at least the >basic functioning of the application and ideally have tested it in a way >that I know will hit the codepath I changed. > >Figuring out how to test a given application, preparing a suitable test >environment etc is often more work than writing the actual patch. That >is why in this case I merely submitted a patch to the bug report and >did not upload as a NMU myself. > >IMO your actions testing that the fix was indeed ready for upload are as >valuable as my figuring out what needed to be changed to make the >package build. absolutely not ;) the testing *is* complete, network tested, and some emails have been migrated successfully from an account to another. My big fear is that (back to my Bachelor/Master Degree in computer science), I did a c program with the above structs, and for a bug I was not filling them correctly. For some implementation reasons (I don't recall the exact code), if the struct is empty, it works correctly, when launched in localhost socket. So I was testing my c program opening a socket on localhost, and connecting to it on the same machine. Everything has been good until I moved the server on another machine :/ The fix has been to fill correctly that structure, and everything has been fixed. So, the test *is* complete, tested with an online imap server, so the network side is fully tested, and mails are transferred between mailboxes correctly (this friend even sent me screenshots, I can share them to you privately to avoid leaks of emails or personal data about his mailserver). But networks are complex, what works for me, and my friend might break on ipv6 networks or whatever else is configure differently. So, in general, even if testing is comprehensive, it won't be 100% complete, regardless on hard you try to check the program. I Hope this clarify my statements in the above email, as a network engineer, I learn on a daily basis that there is no way to reliably check such code (I'm pretty sure it might break/not work on some cntlm proxy, or whatever else, like it will break half of the Debian archive). Hope it is clear now :D thanks a lot for the patch! Gianfranco
