[Bioc-devel] svn and package version numbers

[Dan Tenenbaum]

On Fri, Jul 5, 2013 at 1:40 PM, Kasper Daniel Hansen
<kasperdanielhansen at gmail.com> wrote:
> So what I saw yesterday (I think .. :) is that bumphunter was failing R CMD
> check (it does not fail today) for version 1.1.7 (and earlier versions as
> well) and still I was able to download the windows binary for 1.1.7.  As I
> understand it, that should be impossible since it failed R CMD check.
>

If version 1.1.7 *ever* passed build and check, then it would get propagated.

It seems that whatever causes the failure is transient (doesn't always happen).
So that could cause this situation.
Dan


> Best,
> Kasper
>
>
> On Fri, Jul 5, 2013 at 4:37 PM, Dan Tenenbaum <dtenenba at fhcrc.org> wrote:
>>
>> On Fri, Jul 5, 2013 at 1:33 PM, Kasper Daniel Hansen
>> <kasperdanielhansen at gmail.com> wrote:
>> > My confusion is total.
>> >
>> > Suppose a package gets build (no errors) but fails check (error) and
>> > gets a
>> > version bump.  Does the package source then get propagated to the
>> > web/repository?
>> >
>>
>> I'm confused by your question (sorry).
>> From the point of view of the build system, build and check errors are
>> the same. Both will prevent a package from being propagated to the
>> web/repository.
>>
>> If a package fails check and then gets a version bump it will not
>> propagate to the web, because presumably a mere version bump did not
>> fix the problem that caused the check failure.
>>
>> Dan
>>
>>
>> > Best,
>> > Kasper
>> >
>> >
>> > On Fri, Jul 5, 2013 at 3:29 PM, Dan Tenenbaum <dtenenba at fhcrc.org>
>> > wrote:
>> >>
>> >> On Thu, Jul 4, 2013 at 7:00 PM, Kasper Daniel Hansen
>> >> <kasperdanielhansen at gmail.com> wrote:
>> >> > So this thread is good for my follow up question.
>> >> >
>> >> > I am setting up a Windows box to test/fix bumphunter (newest version
>> >> > 1.1.7)
>> >> > which has been broken (R CMD check) for a while, especially on
>> >> > Windows.
>> >> >
>> >> > This page (bottom)
>> >> >   http://bioconductor.org/packages/2.13/bioc/html/bumphunter.html
>> >> > suggests that the latest binary version for Windows is 1.1.0 which is
>> >> > also
>> >> > probably the last time it build and checked properly.  So this all
>> >> > reflects
>> >> > my expectations of not being able to get a binary version newer than
>> >> > 1.1.0
>> >> >
>> >> > However, when I install R-3.0.1 under windows, source biocLite and do
>> >> > useDevel(TRUE) (getting BiocInstaller version 1.11.3) and then do
>> >> >   biocLite("bumphunter")
>> >> > I get version 1.1.7. Why?  I am asking for binary versions and nor
>> >> > source.
>> >> >
>> >>
>> >> The website wasn't being updated due to an unrelated issue. Fixed now.
>> >> Dan
>> >>
>> >>
>> >> > Best,
>> >> > Kasper
>> >> >
>> >> >
>> >> > On Sun, Jun 30, 2013 at 8:55 PM, Dan Tenenbaum <dtenenba at fhcrc.org>
>> >> > wrote:
>> >> >>
>> >> >>
>> >> >>
>> >> >> On Sunday, June 30, 2013, Kasper Daniel Hansen
>> >> >> <kasperdanielhansen at gmail.com> wrote:
>> >> >> > Doesn't this mean that the issue Wolfgang discusses only arises
>> >> >> > when
>> >> >> > people install from subversion?
>> >> >>
>> >> >> Yes.
>> >> >>
>> >> >> Dan
>> >> >>
>> >> >>
>> >> >> > Kasper
>> >> >> >
>> >> >> > On Sun, Jun 30, 2013 at 6:56 PM, Martin Morgan
>> >> >> > <mtmorgan at fhcrc.org>
>> >> >> > wrote:
>> >> >> >>
>> >> >> >> On 06/30/2013 03:32 PM, Dan Tenenbaum wrote:
>> >> >> >>>
>> >> >> >>> On Jun 30, 2013 12:43 PM, "Kasper Daniel Hansen" <
>> >> >> >>> kasperdanielhansen at gmail.com> wrote:
>> >> >> >>>>
>> >> >> >>>> Also, as far as I understand, the package does not get build
>> >> >> >>>> using
>> >> >> >>>> the new
>> >> >> >>>> commit, if it has already been build with that version number
>> >> >> >>>> before.
>> >> >> >>>
>> >> >> >>> The package is built but not propagated to the web/repository.
>> >> >> >>> This
>> >> >> >>> "feature" allows developers to check that their changes get
>> >> >> >>> built
>> >> >> >>> by
>> >> >> >>> the
>> >> >> >>
>> >> >> >> It's more a safety measure -- if the developer FORGETS to bump,
>> >> >> >> then
>> >> >> >> at
>> >> >> >> least we are not distributing two implementations under the same
>> >> >> >> version
>> >> >> >> number.
>> >> >> >>
>> >> >> >> As Wolfgang says, version numbers are free so no need to hold
>> >> >> >> back
>> >> >> >> on
>> >> >> >> their use.
>> >> >> >>
>> >> >> >> Martin
>> >> >> >>
>> >> >> >>> build system, but yes, once you're satisfied that things work,
>> >> >> >>> you
>> >> >> >>> should
>> >> >> >>> bump the version number to propagate the package and avoid the
>> >> >> >>> confusion
>> >> >> >>> Wolfgang describes.
>> >> >> >>>
>> >> >> >>> Dan
>> >> >> >>>
>> >> >> >>>
>> >> >> >>>>
>> >> >> >>>> Best,
>> >> >> >>>> Kasper
>> >> >> >>>>
>> >> >> >>>>
>> >> >> >>>> On Sun, Jun 30, 2013 at 3:12 AM, Wolfgang Huber
>> >> >> >>>> <whuber at embl.de>
>> >> >> >>>> wrote:
>> >> >> >>>>
>> >> >> >>>>> Hi All,
>> >> >> >>>>>
>> >> >> >>>>> just a reminder that it is good practice to bump up the
>> >> >> >>>>> package
>> >> >> >>>>> version
>> >> >> >>>>> when you commit a change to a package's source, even if you
>> >> >> >>>>> consider
>> >> >> >>>>> it
>> >> >> >>>>> 'trivial'. Version numbers are free, while the confusion
>> >> >> >>>>> ensuing
>> >> >> >>>>> from
>> >> >> >>>
>> >> >> >>> there
>> >> >> >>>>>
>> >> >> >>>>> being different versions of the software with ostensibly the
>> >> >> >>>>> same
>> >> >> >>>
>> >> >> >>> version
>> >> >> >>>>>
>> >> >> >>>>> can waste a great deal of someone's time.
>> >> >> >>>>>
>> >> >> >>>>> Dan / Bioc-Core team: would it be good to mention this
>> >> >> >>>>> somewhere
>> >> >> >>>>> on
>> >> >> >>>>> http://bioconductor.org/developers/source-control ?
>> >> >> >>>>>
>> >> >> >>>>> Best wishes
>> >> >> >>>>>          Wolfgang
>> >> >> >>>>> _______________________________________________
>> >> >> >>>>> Bioc-devel at r-project.org mailing list
>> >> >> >>>>> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>> >> >> >>>>>
>> >> >> >>>>
>> >> >> >>>>          [[alternative HTML version deleted]]
>> >> >> >>>>
>> >> >> >>>> _______________________________________________
>> >> >> >>>> Bioc-devel at r-project.org mailing list
>> >> >> >>>> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>> >> >> >>>
>> >> >> >>>         [[alternative HTML version deleted]]
>> >> >> >>>
>> >> >> >>> _______________________________________________
>> >> >> >>> Bioc-devel at r-project.org mailing list
>> >> >> >>> https://stat.ethz.ch/mailman/listinfo/bioc-devel
>> >> >> >>>
>> >> >> >>
>> >> >> >>
>> >> >> >> --
>> >> >> >> Computational Biology / Fred Hutchinson Cancer Research Center
>> >> >> >> 1100 Fairview Ave. N.
>> >> >> >> PO Box 19024 Seattle, WA 98109
>> >> >> >>
>> >> >> >> Location: Arnold Building M1 B861
>> >> >> >> Phone: (206) 667-2793
>> >> >> >
>> >> >> >
>> >> >
>> >> >
>> >
>> >
>
>