hallo Lars,

Lars Schimmer wrote:
> For windows only snapshots and unofficial build >0.7.0 are noted, but
> the windows bin link follows to http://hugin.huikeshoven.org/ which
> tells me "only vista, highly experimental" for >0.8.0.
> 
> Is there no stable (general) windows binary release >0.7.0 available at all?

Nicolas gave you the short answer: No. With a little bit of fiddling you 
can make the experimental builds work for you, but they are not of the 
same stability / quality as the 0.7.0 and the critical criteria to be 
uploaded to SF and distributed as official is real stability, not just 
"declared stability" (as in: the project has released them).

Long answer: There are four issues that make a Windows binary installer 
complex. Build-time dependencies; Run-time dependencies; the Installer 
itself; and last but not least the SDK, that brings them all together.

A Windows binary release must also ship other run-time dependencies 
because of the lack of proper package management system on that 
platform. On platforms with a package manager, the four issues are taken 
automatically care of..

Around the time of the 0.7.0 we had a lucky combination of really stable 
dependencies; up to date Installer scripts; and users from the Windows 
community willing to go through the pains of a systematic and clean 
build (that was myself and Guido. In the end his binaries were uploaded 
to SF). Since then things have been wacky. They may improve by the end 
of the year.

Current status:

BUILD-TIME DEPENDENCIES

1. Libpano is the critical build-time dependency. Bleeding edge features 
require the latest libpano. The current 2.9.15_beta3 is acceptable 
(although there have been further fixes committed, so better wait for 
the next tarball). In between the repository was broken and either does 
not build or may result in faulty Hugin binaries.

2. Other build time dependencies don't improve as fast; tracking their 
improvements is not critical. The current SDK (by Guido) should be good 
enough to build Hugin. Using slightly older dependencies only affects 
marginally the end-user experience and is acceptable.


RUN-TIME DEPENDENCIES

There are two critical run-time dependencies: Enblend, and (optional) a 
control point generator.

1. Enblend is the critical run-time dependency. At the time of releasing 
Hugin 0.7.0 we were lucky that the Enblend repository was in a stable 
state. After that, Enblend 3.2 has been released and there are reports 
that it is a regression compared to the pre 3.2 snapshots. To compile an 
  Hugin installer Enblend must be present inside the SDK. Unfortunately 
the most recent SDK does not have what it take to build Enblend. Joachim 
is expanding the SDK to build Enblend from the repository. Enblend is 
now in the final stages of polishing / cleaning up bugs ahead of the 4.0 
release. There is no guaranteed release date, the stated objective is to 
release before year end. No guarantee.

2. There are a bunch of options for the control point generator. The 
problem is that all of those that work are covered by some patent (SIFT 
or SURF) in some jurisdictions, making their distribution a hot potato. 
Although a control point generator is an optional dependency, most 
end-user consider it mandatory. The script to compile the 0.7.0 
installer had two variations, one with and one without CP generators. 
I've expressed the view that in the future it would be good to do the 
same as for OSX and release separate Windows binary installers for Hugin 
and for the CP generators.

3. Autopano-SIFT-C is the last incarnation of the CP generator 
traditionally associated with Hugin. The current status is that the 
previous "stable" release (2.5.0) had serious memory leaks and crashed 
on large projects. The current development snapshot of 2.5.2 is broken, 
and there is no guaranteed date for the fix (Tom said he has no time 
before the end of the year). As an interim solution we've branched out 
2.5.1 (currently release candidate 2, could be declared final any time).

4. Autopano-SIFT-C is covered by the SIFT patent in some countries. The 
patent holder, UBC, allows use for "educational purposes". I do not 
recall if also for "non-commercial purposes". The patent is the reason 
why popular mainstream Linux distributions such as Debian and Fedora 
don't carry Autopano-SIFT-C.

5. A better control points generator is Pano-O-matic. It is covered by 
the SURF patent. You can find it on the web, but it is very unlikely 
that it will make it into the official Hugin binary distribution. You 
can make it work for you if you know what you are doing.

6. We still don't have a patent-free, open sourced control points 
generator. Some parts have been produced during Google Summer of Code 
2007 and 2008, but work still needs to be done and there is nobody 
working on that at the moment, AFAIK.


WINDOWS INSTALLER

Since I added the 0.7.0 installer scripts to Hugin's codebase I changed 
platform and the scripts were practically orphaned. All installers 
circulating after 0.7.0 are modified versions and until not long ago the 
modifications where not even back in the codebase. Allard has committed 
his modified version recently. It has not been tested/verified by 
anybody else.

All of these issues can be solved. Some of them are related to the 
Hugin/Enblend/Libpano/Autopano-SIFT-C codebase. We're solving these. The 
new parallel development cycle of Hugin works well, trunk is in better 
state than ever and we can branch out regularly quite stable release 
codelines. Libpano seems to be going a similar way, with 2.9.15 branched 
out as a separate codeline. This bodes well for future stability. Bruno 
stabilized an Autopano-SIFT-C interim release and Chris is taking care 
of Enblend.

Some of them are platform specific and are left to volunteers on that 
platform. Harry has been the driving force behind OSX. We have not had a 
driving force behind Windows since 0.7.0. Some Windows users are 
contributing valuable code (Thomas), build (Allard, Ad, Joachim), SDK 
(Guido, Ryan), and testing (too many to name, and sorry if I forgot 
somebody contributing code, build, SDK work). But nobody is really 
taking it all the way to a proper and clean binary release. It's a lot 
of work, not to be underestimated. It takes more time than just syncing 
the repositories and building; and for those like me who have access to 
multiple platforms, it is time taken from other priorities.

I hope this is shed some light on the situation...

> - -------------------------------------------------------------
> TU Graz, Institut für ComputerGraphik & WissensVisualisierung

... at such an institute you should have students / assistants / 
professors with the skill to build Windows code?

Yuv

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"hugin and other free panoramic software" group.
A list of frequently asked questions is available at: 
http://wiki.panotools.org/Hugin_FAQ
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/hugin-ptx
-~----------~----~----~----~------~----~------~--~---

Reply via email to