Build ID: > Mozilla/5.0 (Windows NT 5.1; rv:9.0a1) Gecko/20110828 Thunderbird/9.0a1 Detatch/Delete at "All Body Parts" works perfectly - any part under multipart/xxxx can be detached/deleted according to user's request. However, "Detach All/Delete All" at "View/Messae Body As/All Body Parts"(introduced by Bug 602718) is too dangerous for general users, because any part including message body part(text/html usually), referred/embed part(image pointed by cid: usually), is detached/deleted. I believe that at least "Detach All"/"Delete All" at "View/Messae Body As/All Body Parts" should be disabled, in order to avoid many unwanted bug reports. I think that Detach/Delete at "View/Messae Body As/All Body Parts" is better disabled, although many bugs for complaints like bug 351224 comment #0 will probably be opened if detach/delete will be disabled.

WTF? First: new functionality can't regress Second: why was this added if it's too 'dangerous'? Third: just add a confirmation dialog

mail-1 : Original multipart/related mail mail-2 ; Mail after Detach All of mail-1 mail-3 ; Mail after Delete All of mail-1

(In reply to Ferry Huberts from comment #1) > First: new functionality can't regress Sorry, my mistake. I initially created this bug as "clone of Bug 672718", so keyword, dependency tree was copied from that bug. > Second: why was this added if it's too 'dangerous'? Detech/delete was not added by enhancement of Bug 672718. Design/implementation of detach/delete is unchanged. I guess next; After sort out/clean up by Bug 351224 and ehancement by Bug 672718, detech/delete works as desinged on any part which is shown as(or as if) atachment at attachment pane.

Sorry typo. Bug 672718 -> Bug 602718 in previous comment.

Let's keep in mind that the fix in bug 602718 has to be explicitly enabled by the user to even see that menu point, thus "too dangerous for a general user" doesn't apply in this case given that the option isn't available without intervention by the user toggling a hidden preference. Thus, this case is well beyond general use and shouldn't pose any threat of accidental deletion to an inexperienced user. Of course, this assumes that someone who is explicitly pointed to that option (e.g., in a forum thread) is informed about any quirks possible in this mode. Personally I'd prefer a solid solution in bug 674473 in order to limit the cases where actually using the hidden option becomes necessary rather than introducing yet as special case here to catch potentially harmful operations in this mode.

How hard would it be to use the new mime code when doing detach/delete of attachments, and the old code when doing message display, as a somewhat temporary fix?

the previous comment should have been in bug 351224 and the regression it introduced.