Who has commit rights? Use it!
Hi, I recently posted a patch which fixes some g++ 4.3 compiler errors and warnings but it was mostly ignored. I was forced that's why to create bug reports (which you call ticket, don't know why) for each library (which is time consuming especially without email interface) (#1332 -- #1341). I assumed trivial bugs such as Index: boost/gil/extension/dynamic_image/apply_operation_base.hpp =================================================================== --- boost/gil/extension/dynamic_image/apply_operation_base.hpp (Revision 40416) +++ boost/gil/extension/dynamic_image/apply_operation_base.hpp (Arbeitskopie) @@ -100,7 +100,7 @@ // Create specializations of apply_operation_fn for each N 0..100 GIL_GENERATE_APPLY_FWD_OPS(99) -}; +} // unary application template <typename Types, typename Bits, typename Op> which lead to real compile errors (at least using -pedantic) will be fixed during minutes but I was wrong! Of course I know that not everyone can react as fast but why do not commit other people such elementary stuff? Instead many simple patches are rotting in the bug database. Is this useful? Is it enforced somewhere that only the maintainer is allowed to commit patches for his library? I understand of course that not trivial stuff which needs to be analysed cannot be handled this way but there are so many trivial patches just ignored ... Jens
Is it enforced somewhere that only the maintainer is allowed to commit patches for his library? Not enforced, but seen as good manners. Would you like it if someone would, say, while you're cooking, constantly reach in and do minor tasks for you, unasked? I would be annoyed, because even with minor tasks,
Jens Seidel wrote: there's a chance of something going wrong. Also, there's the problem of noticing. If you submit a ticket (it's called ticket because that's what the system calls it) for library XYZ, then only the maintainer of XYZ is really aware of it. Other people have better things to do than go through patches for libraries other than their own looking for patches trivial enough that they can apply them. Also, GCC 4.3 is, at the moment, still a fuzzy target. Things can change. That doesn't apply to the example patch, of course (although I had to check the standard to realize that a lonely semicolon really isn't allowed on the global level), but in general, people will be wary when confronted with fixes for an experimental compiler. Sebastian Redl
Jens Seidel wrote:
Hi,
I recently posted a patch which fixes some g++ 4.3 compiler errors and warnings but it was mostly ignored. I was forced that's why to create bug reports (which you call ticket, don't know why) for each library (which is time consuming especially without email interface) (#1332 -- #1341).
Have you considered how much unpaid time some of the people here already spend in maintaining Boost?
I assumed trivial bugs such as
Index: boost/gil/extension/dynamic_image/apply_operation_base.hpp =================================================================== --- boost/gil/extension/dynamic_image/apply_operation_base.hpp (Revision 40416) +++ boost/gil/extension/dynamic_image/apply_operation_base.hpp (Arbeitskopie) @@ -100,7 +100,7 @@
// Create specializations of apply_operation_fn for each N 0..100 GIL_GENERATE_APPLY_FWD_OPS(99) -}; +}
// unary application template <typename Types, typename Bits, typename Op>
which lead to real compile errors (at least using -pedantic) will be fixed during minutes but I was wrong!
I don't know any *paid* support service that guarantees this kind of response. Your problem may look trivial but anyone taking charge of it would have to actually check that your patch is meaningful, apply it locally, run the library tests on a few platforms, including the one against which the ticket was filed. Add the fact that we're talking about an unreleased compiler version, which means that it is not one of the release critical platforms and that it's probably not so commonly used among the Boost maintainers; I would be surprised if your problem was taken care of as fast as you would like.
Of course I know that not everyone can react as fast but why do not commit other people such elementary stuff? Instead many simple patches are rotting in the bug database. Is this useful?
Is it enforced somewhere that only the maintainer is allowed to commit patches for his library? I understand of course that not trivial stuff which needs to be analysed cannot be handled this way but there are so many trivial patches just ignored ...
No, the usual practice is for someone with commit rights to announce that they have a patch for a specific library and then wait a few days. If no objection comes the patch is usually applied, with the understanding that the committer will revert it should it cause unexpected problems. Cheers, Nicola -- Nicola.Musatti <at> gmail <dot> com Home: http://nicola.musatti.googlepages.com/home Blog: http://wthwdik.wordpress.com/
participants (3)
-
Jens Seidel -
Nicola Musatti -
Sebastian Redl