Hugo and Thomas,

In the light of day I want to make it clear what I tried to say 
in the early hours (of my long standing insomnia):

First, thank you, Thomas for coming into the discussion and 
thank you Hugo for taking up the matter.  Further comments 
inside the text:
On Wednesday 29 December 2010 01:04:40 Thomas Perl wrote:
> Hi all,
> 
> 2010/12/28 Hugo Parente Lima <[email protected]>:
> > On Tuesday 28 December 2010 00:36:46 Algis Kabaila wrote:
> >> On 2010-11-19 Matti Airas wrote:
> >> """
> >> We now have (in the yet-unreleased Git head version)
> >> version strings as such:
> >> 
> >> PySide version: PySide.__version__ and
> >> PySide.__version_info__ Qt version:
> >> PySide.QtCore.__version__ and
> >> PySide.QtCore.__version_info__
> >> """
> >> This version information is appreciated.  However, it
> >> would be even more useful if it displayed the minor
> >> version information, perhaps "b1" for beta1 and "b2" for
> >> beta 2 so that it would be possible to see what version
> >> one is operating.
> >> 
> >> As the little snipped of program below will show,
> >> currently for both b1 and b2 the above strings display
> >> the following: PySide.__version__   shows  string "1.0.0"
> >>  and
> >> PySide.__version_info__ shows string "(1,0,0)"
> > 
> > Yes... this is an issue, we can append "~betaX" on
> > __version__ but I don't know how to proceed with
> > __version_info__, because it's a tuple of ints besides we
> > don't want this tuple to have different sizes on different
> > PySide versions.
> 
> The 1.0.0 beta 1 could have a version number of 0.99.1 (beta
> 2 could have 0.99.2, etc..) - this would allow the version
> number to always be "dotted decimal" and have the versions
> correctly ordered by comparison.
> 
> I know that we can't change this for now, but maybe this
> could be a consideration for the future (i.e. the "1.1" beta
> 1 could be "1.0.99.1" or whatever).
> 
> This way, beta versions are sorted "before" the final
> releases, but still work fine for total ordering. Let's say
> the imaginary version 1.1 would have four beta releases and
> the final release:
> 
> 1.0.99.1
> 1.0.99.2
> 1.0.99.3
> 1.0.99.4
> 1.1

The immediately above is unnecessarily confusing and breaks the 
requirement of not counting backwards version numbers.  
Acceptible,  however, would be:

(1, 0, 0, 1)
(1, 0, 0, 2)
etc
up to
(1, 0, 1)

Why would you, Thomas,  suggest to jump and leave a gap and 
instead of 
(1, 0 , 1) 
write 1.1 is not clear to me.  BTW, I assume you meant to write 
the above as tupples and not strange "numbers" with several "." 
separators

AFAICS (As Far As I Can See) it would also be fine to leave the 
PySide.__version_info__  as (1, 0, 0) until it becomes 
(1, 0, 1).

I think we all agree about PySide.__version__  -- the two last 
versions should have been either
"1.0.0-bX" or "1.0.0-betaX" with X in the range of (1, 2) with a 
proviso that it may not be practical to go back in time to have 
"1.0.0-b1".

It would in MHO be perfectly OK to start with "1.0.0-b2"  or 
"1.0.0-beta2".  It is perfectly acceptable to jump forward the 
version numbers (but it is very confusing to go backward in 
version numbers and thus un-acceptable)

The outline of this supplementary reply is in line with past 
practices - and that is IMO a necessary prerequisite.

> 
> Now, you write your application and depend on a feature
> introduced in beta 2 (i.e. "1.0.99.2") - you could now make
> a version check for "greater than 1.0.99.2", and it will
> have the correct dependency (assuming that the feature isn't
> removed in later betas). Oh, and due to the way tuples are
> compared, you could literally write "PySide.__version_info__
> >= (1, 0, 99, 2)" :)
> 
> Again, this does not solve the current problem, but it could
> help in defining a policy to follow for future (post-1.0.0)
> releases and beta releases :)
> 
> 
> HTH.
> Thomas

In summary, suggest to change the in the current Head  
PySide.__version__ = "1.0.0.b2"
and
PySide.__version_info__ = (1, 0, 0, 2)  or (1, 0, 0) if a three 
element tuple is mandatory.


With kind regards and best wishes for the coming New Year

Al.

-- 
Algis
http://akabaila.pcug.org.au
_______________________________________________
PySide mailing list
[email protected]
http://lists.openbossa.org/listinfo/pyside

Reply via email to