Return to stuff_we_dislike_about_gnats.mdwn CVS log

Up to [NetBSD Developer Wiki] / wikisrc / bugtracking

Add bsiegert's wish for more meaningful states of "closed".

    1: # Stuff we dislike about gnats
    2: 
    3: ## Storage and filing
    4: 
    5: * gnats does not understand MIME attachments. (At all.) They end up as
    6: text blobs in the database. And worse, they end up mildly corrupted so
    7: they need to be edited to be decoded.
    8: 
    9: * gnats makes it a nuisance to extract even inline patches from PRs.
   10: 
   11: * gnats has no way to prevent accidentally filing messages in old PRs,
   12: and it's frequently necessary to hand-edit old PRs to clean up the
   13: consequences.
   14: 
   15: * gnats accepts a wide range of misformatted incoming PRs, then
   16: mangles them further and drops them into an internal queue needing
   17: manual attention.
   18: 
   19: * gnats fairly frequently belches on incoming mail requiring
   20: intervention from admins.
   21: 
   22: * By default gnats does not treat incoming confidential PRs as
   23: confidential; it requires manual filtering in the surrounding email
   24: logic.
   25: 
   26: * gnats has no support whatsoever for automatically handling bounced
   27: mail.
   28: 
   29: * There's no way to disable sending mail to email addresses that are
   30: known to be invalid.
   31: 
   32: * When a bug is closed, there's no way to disambiguate from its
   33:   closed state whether it's fixed, invalid, obsolete, or otherwise
   34:   "won't fix".
   35: 
   36: ## User interface
   37: 
   38: * gnats has no ability to do anything by web besides display PRs, and
   39: they display only as plain text. Things that you should be able to do
   40: that you can't include:
   41:    * filing comments
   42:    * browsing from one PR to other related ones (either explicit links
   43: from one PR to another, or retriving other PRs with similar
   44: properties)
   45:    * subscribing to PRs to get updates (even by email, let alone by RSS)
   46: 
   47: * The only way to do administrative operations is to run an editor on
   48: the backend text file representation of a single bug. This means you
   49: can't readily:
   50:    * massedit PRs (e.g. put a lot of related PRs in feedback at once)
   51:    * update someone's email address
   52:    * move misfiled comments from one PR to another
   53: 
   54: * But you can't file a comment on a PR from the command line; you have
   55: to send email.
   56: 
   57: * The nag mails that gnats sends to submitters are annoying to handle;
   58: in particular just replying to them does not go anywhere useful.
   59: 
   60: * The nag mails that gnats sends to developers list a bunch of PRs but
   61: do not list them in any kind of useful order; nor does gnats note
   62: what's different since the previous one.
   63: 
   64: ## Querying and search
   65: 
   66: * There's a web search form, but it's close to useless, partly because
   67: it doesn't take input well and partly because the lists of results it
   68: produces is unsorted, unstructured, and impossible to work with.
   69: 
   70: * The command-line interface supports search... using a custom query
   71: "language" that's nearly impossible to use and that furthermore has
   72: most of the same drawbacks as the web search page. In order to do
   73: anything much useful you need to write your own scripts that wrap the
   74: output.
   75: 
   76: * Finding stuff in the database is sufficiently hard that many people
   77: consider it effectively impossible.