Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/~mperrin/software/gpidata/developers/releases_and_versioning.html
Дата изменения: Sat Feb 15 03:42:02 2014
Дата индексирования: Sun Mar 2 12:23:18 2014
Кодировка: IBM-866

Поисковые слова: arp 220
Release Guide: Versions, Version Tracking, etc — GPI Data Pipeline 1.0 documentation

Table Of Contents

Previous topic

Editing Documentation (Including this Website)

Next topic

Appendix: Recipe XML File (‘DRF’) Format

This Page

Release Guide: Versions, Version Tracking, etc€ґ

Warning

This page contains working notes for making new release copies of the pipeline.

This page is not likely to be of much use to anyone other than members of the GPI team.

Branches€ґ

Most pipeline development happens in the trunk directory. Right now we have one branch directory, called public. This allows the GPI team to have internal development routines that aren’t included in the release zip files. (Note: the intent here is not to keep secrets from anybody, it’s to avoid shipping incomplete/experimental/known-broken code in the public releases. These new functions will generally make their way into public once they’re working properly.)

Creating a New Release Version€ґ

This is a methodical process with lots of steps - eventually there will be a script to automate this better. Once your pipeline codebase is ready to release:

The first step in making a new release version is to increment the version number. This needs to be changed in two places.

shell>  vi backbone/gpi_pipeline_version.pro
shell>  vi documentation/conf.py
shell>  svn commit backbone/gpi_pipeline_version.pro documentation/conf.py -m "increased version number to XX.YY"

Even though zip files do not yet exist at this point, you may wish to add placeholder links to the downloads pages.

shell>  emacs documentation/install/install_from_zips.rst
shell>  emacs documentation/install/install_from_source.rst
shell>  svn commit documentation -m "add download links for version XX.YY"

Then update the public branch:

shell> cd repository/pipeline/branches/public
shell> svn merge https://repos.seti.org/gpi/pipeline/trunk

You may possibly have to deal with and resolve conflicts if there are any at this point. Hopefully not. Any files that should be excluded from the public release can be removed from the public branch at this point. Then commit the public branch:

shell> svn commit -m "Merged changes from trunk to public branch"

The next step is to tag the released version in subversion. This is done by svn copying the relevant trunk directories into appropriate tags directories, under both pipeline and external:

setenv VER 0.x.y
svn copy https://repos.seti.org/gpi/pipeline/branches/public https://repos.seti.org/gpi/pipeline/tags/${VER} -m "Release copy of pipeline version ${VER}"
svn copy https://repos.seti.org/gpi/external/trunk https://repos.seti.org/gpi/external/tags/${VER} -m "Release copy of pipeline external dependencies version ${VER}"

This does the update on the server only. To update your local copy of the release directory, you can do an svn update in your tags directory.

To create the compiled executables (‘sav files’ in IDL speak), run the gpi_compiler routine in IDL, and when prompted enter the desired output directory. Before starting IDL, you should adjust your IDL $IDL_PATH so that it points at the public directory, and is free of any of your personal routines or other IDL code. This will ensure that you’re compiling the versions in the pipeline directory. You should also make sure that your $IDL_PATH is pointing toward the public directory:

shell ~ > setenv IDL_PATH "+/home/username/public_pipeline:+/home/username/external:+/Applications/itt/idl/idl81/lib"
IDL> gpi_compiler
Enter name of the directory where to create executables:  ~/tmp/gpi
[very long series of informative messages during compilation]
% GPI_COMPILER: Compilation done.
% GPI_COMPILER: The ZIP files ready for distribution are:
% GPI_COMPILER:       ~/tmp//gpi_pipeline_0.9.3_r2345M_runtime_macosx.zip
% GPI_COMPILER:       ~/tmp//gpi_pipeline_0.9.3_r2345M_compiled.zip
% GPI_COMPILER:       ~/tmp//gpi_pipeline_0.9.3_r2345M_source.zip

If you are compiling on Mac or Linux, gpi_compiler will automatically zip up the output into three zip files:

  • a platform independent one containing just the save files
  • and an OS dependent one that contains also the IDL runtime virtual machine
  • and a matching source code ZIP file as well.

Note

It is normal to see a handful of compiler error messages when building the pipeline. Generally these are scripts or code blocks that get sourced into other routines, that are not intended to be compiled on their own. The gpi_compiler infrastructure ought to be updated to be smarter about ignoring these, but in the mean time these errors can just be ignored.

Upload the resulting zip files to the desired download locations. Update the documentation source to have the proper ZIP file locations, if needed, and recompile using Sphinx. Email Franck to update the official documentation on `http://docs.planetimager.org/pipeline`_

Switching to a given release on subversion€ґ

In your working copy of the ‘pipeline‘ directory, for instance:

svn switch https://repos.seti.org/gpi/pipeline/tags/0.9.2/

To check this has taken effect:

svn info

and check the URL line in the output has the release tag in it.