Differences between revisions 14 and 15
Revision 14 as of 2016-05-03 15:25:07
Size: 10477
Editor: was
Comment:
Revision 15 as of 2016-05-04 12:52:27
Size: 11398
Editor: egourgoulhon
Comment:
Deletions are marked like this. Additions are marked like this.
Line 14: Line 14:
combinatorics on words, Sage-manifolds for differential geometry, combinatorics on words, SageManifolds for differential geometry,
Line 108: Line 108:

Examples:

- `SageManifolds <http://sagemanifolds.obspm.fr/>`_, cf. the metaticket `#18528 <http://trac.sagemath.org/ticket/18528>`_
Line 173: Line 177:
- `Template for creating Sage packages <https://github.com/cswiercz/sage_packages>`
- `Modular Abelian Varities <https://github.com/williamstein/sage_modabvar>`
- `Python implementation of chebfun <https://github.com/cswiercz/pychebfun>`
- `Purple Sage <https://github.com/williamstein/psage>`
- `Template for creating Sage packages <https://github.com/cswiercz/sage_packages>`_
- `Modular Abelian Varities <https://github.com/williamstein/sage_modabvar>`_
- `Python implementation of chebfun <https://github.com/cswiercz/pychebfun>`_
- `Purple Sage <https://github.com/williamstein/psage>`_
Line 183: Line 187:
- `Sage-Manifolds <http://sagemanifolds.obspm.fr/>`: the install script (http://sagemanifolds.obspm.fr/spkg/sm-install.sh) for this does all kinds of copying files directly into the sage install, using sed to modify parts of the sage library, etc. It's terrifying. -- William - `SageManifolds <http://sagemanifolds.obspm.fr/>`_: the install script (http://sagemanifolds.obspm.fr/spkg/sm-install.sh) for this does all kinds of copying files directly into the sage install, using sed to modify parts of the sage library, etc. It's terrifying. -- William. True but that's only a provisory thing for an easy install by a newbie, until the process of full integration of SageManifolds in Sage, started at `#18528 <http://trac.sagemath.org/ticket/18528>`_, is finished. For a developer, the recommended installation process is via git, not via the above script. Actually, we started SageManifolds as a (new-style) spkg and it was distributed as such until version 0.4. Then, in order to ease the review process, we split it in many tickets (listed at `#18528 <http://trac.sagemath.org/ticket/18528>`_) and devise the above script just for end users not familiar with git and make. Really the development workflow of SageManifolds pertains to the category ''Direct integration into Sage'' above, hence it is a Non-Example here -- Eric.

On development workflows for sharing (experimental) code

One core aim of Sage is to foster code sharing, and encourage groups of researchers, teachers, and other users to get together to develop new features they need either on top of or within Sage, and share them.

Over the years, many development workflows have been experimented by various groups of people to improve Sage in certain areas, like Sage-Combinat for (algebraic) combinatorics, Sage-Words for combinatorics on words, SageManifolds for differential geometry, purple-sage for number theory, ...

The goal of this document is to discuss the different workflow that have been tried, their pros and cons, to share best practices and brainstorm about what support and recommendations Sage could provide for various use cases.

At this point this is a collection of notes by N. Thiéry; please hack in and contribute your own vision!

Objectives of a development workflow

Of course the milleage will vary from project to project, but the objectives of a development workflow can typically be to:

  1. Support fast paced development within a group of users working on the same topic, or needing similar features.
  2. Support rapid dissemination of experimental features

    The goal is simultaneously to support users, and to get early feedback on the code.

    Typical needs:

    • Using, for a given calculation, experimental features from different areas, developped by different groups of people
    • Getting the latest version of a feature, without having to upgrade all of Sage (e.g. just before delivering a talk!!!)
    • Feature discovery: increasing the chances for someone to discover that a given feature is being implemented somewhere.
  3. Foster high quality code by promoting documentation, tests, code reviews
  4. Foster intrinsic high quality code by providing an ecosystem where (experimental) code can live, compete with other implementations, mature and be selected or die, all at a fine granularity.
  5. Strike a balance between centralized and decentralized.

    In particular mitigate the risks of code-bloat of the Sage library versus the risks of death of code lying out somewhere on the web.

  6. Minimize maintenance overhead, and in particular code rotting
  7. Remain flexible between the all-in-one versus packages development models (simplifying things out: between Sage's model and GAP's model)
  8. Promote extending existing Sage classes and modules with additional features.

    This eases dynamic feature discovery by users (once installed, features can be found at their expected place) and enable transparent migration of code inside the Sage library if and when desired (no need to change the code itself, nor code using it). This also promotes coherent coding standards.

    Note: subclassing is not always an option to extend a class, e.g. when a feature is to be added to an abstract base class of many concrete classes (subclassing each and every concrete class would be a pain)

    See also:

Existing workflows

Direct integration into Sage

In this workflow, each feature is shared by integrating it directly into Sage.

Pros:

  • Simplicity for the user: all stable features are directly available in Sage
  • Simplicity for Sage developers: no additional workflow to learn
  • No need to worry about release, distribution, test infrastructure, ...
  • Promotes early integration of code and 3.
    1. is straightforward

Cons:

  • Limited support for 2.
  • Slows down the development: once a feature is in Sage, any change needs to be reviewed, refactoring of the public API requires taking care of backward compatibility. No good for 4.
  • Getting the latest feature forces updating to the latest version of Sage
  • Introduces a bias toward code bloat (in doubt, features tend to be added to Sage)

Examples:

Experimental feature branches

In this workflow, experimental feature or feature sets are implemented as branches on the Sage sources.

Pros:

    1. is straightforward
  • Encourages integration into Sage
  • Development history is automatically kept upon integration into Sage

Cons:

  • Branch needs to be regularly updated to prevent code rotting due to syntactical conflicts with changes in Sage (though automatic merges help).
    1. requires basic git knowledge from end-users.
  • Lack of modularity for 2.: due to potential conflicts, it's not easy to combine features from several branches; upgrading to the latest version of a branch often forces a change of version of Sage
  • Cherry picking certain mature features for integration in Sage is somewhat cumbersome (the granularity of branches and commits is orthogonal to the granularity of features).
  • It's hard to strike the right granularity in terms of feature / feature set. We tried dependency tracking among branches as a way to build feature sets out of features, but this did not work well.
  • Because of the above, this workflow does not work well for 4.
  • Introduces a bias toward the all-in-one development model.

Patch queue as used by Sage-Combinat between 2009 and 2013

See also those old design notes about the Sage-Combinat workflow.

TODO: description

Pros:

  • Relatively good for 1. (except for 6.)
  • Relatively good for 2. (thanks to "sage -combinat install"), except for modularity and requiring some Sage recompilation
    1. is straightforward

Cons:

  • Complexity of working at the meta level (version control on the patches)
  • Really bad at 6: Horrible maintenance overhead due to syntactic conflicts and lack of automatic merging
  • Introduces a strong bias toward code death, or at least non integration into Sage
  • Monolithic: one could not use several patch queues at once, so this did not support overlaping groups of people working on different topics; this introduced a non-natural barrier between Sage-Combinat and the rest of the world, and prevented rapid reconfiguration of projects around topics and groups of developers

Standalone (pip) packages

Here the idea is to implement feature sets as independent Python packages on top of Sage. Converting a bunch of Python files into such a package to make it easy to install is straightforward.

Examples:

NON-Examples:

  • SageManifolds: the install script (http://sagemanifolds.obspm.fr/spkg/sm-install.sh) for this does all kinds of copying files directly into the sage install, using sed to modify parts of the sage library, etc. It's terrifying. -- William. True but that's only a provisory thing for an easy install by a newbie, until the process of full integration of SageManifolds in Sage, started at #18528, is finished. For a developer, the recommended installation process is via git, not via the above script. Actually, we started SageManifolds as a (new-style) spkg and it was distributed as such until version 0.4. Then, in order to ease the review process, we split it in many tickets (listed at #18528) and devise the above script just for end users not familiar with git and make. Really the development workflow of SageManifolds pertains to the category ''Direct integration into Sage'' above, hence it is a Non-Example here -- Eric.
  • CHA "It is recommended to use the more recent implementation from the branch attached to this ticket rather than this library."; I think this is just some code to copy into the sage library or run directly, with no package support at all.

Pros:

  • Good for 1., 2., 4.,

Cons:

  • Handling of compatibility with various versions of the dependencies (in particular Sage)
  • Risk of code rotting (as Sage evolves over time) or death (if it's not maintained)
  • Requires coordination with Sage and related packages to not step on each other

Standalone (pip) packages with an integration mission

This is a variant on the previous development workflow, with an explicit focus on easing (or even promoting) the integration of mature code into Sage.

Specifics:

  • Layout the code as in the Sage library, with top module called e.g. sage-blah instead of sage. For example, to add a method to the Sage class Partition, one would put it in an otherwise empty class sage-blah.combinat.partition.Partition.

  • Use recursive monkey patching (TODO: make a pip package for this, and add a link here) to insert all the code dynamically in the Sage library.

    The effect is to patch the Sage library, as with branches or patch queues; however this is done semantically at the granularity of methods rather than syntactically at the granularity of lines in the source code.

Examples:

Pros:

  • Same as above
    1. is straightforward
  • Lighter maintenance overhead compared to branches or patch queues: one only needs to take care of semantic conflicts, not syntactic ones.
  • The integration of mature code into Sage helps for 3 and for the maintenance as well: keeping the library as a "small layer" over Sage reduces the risks of irreversibly drifting away, and reduces the amount of updating.
  • Depending on how strongly one pushes toward the integration of mature code, one can flexibly interpolate between the all-in-one model and the package model

Cons:

  • The concept has not yet been really battlefield tested!
  • Moving code into the Sage library is done by copy pasting. This makes for a clean diff showing just the addition of the new methods, but means that one looses the history and author tracking (that's not that different from history squashing as used by many projects)

CodeSharingWorkflow (last edited 2023-02-23 21:49:01 by mkoeppe)