- Boost - lists.boost.org

[build] improving the history
by Vladimir Prus 18 Dec '13

18 Dec '13

I've spent some time looking at history of Boost.Build git repository, and I think I've got a fairly reasonable outcome. The problem statement is that in the converted repository, the top-level has nothing but a directory called 'v2', which is rather ugly. While it can be fixed with 'git mv', it would mean that 'git log file' will not show complete history unless the --follow option is passed, which is inconvenient. It appears that graphical tools don't like moves much either. There were already moves in SVN, so even now, running 'git log' on any file in engine/ won't return complete history without --follow. Further, at some older revision, there are two copies of what was jam code then. The most convenient solution for these problems is to edit history of the 'develop' branch, so that content of top-level directory matches what was in 'v2' in SVN, and so that content of 'engine' matches Jam sources through the years. This won't actually break any important references from the super-project, first because they presumably are only to master branch, and second, because the commit objects will be still retained in repository. I am not sure we need to rewrite the 'master' branch, as it's not used for day to day development. Now, as soon as we're rewriting history, it might be good idea to change directory structure to be more alike to the structure that other components have. The additional downside here is that the historical versions will not be fully functional - specifically it will break references from examples and tests to the rest of Boost.Build. But, I think it's worth the trouble - users can still obtain old, fully functional, versions from the tags, or master branch (yet another reason for it not to be rewritten), while developers will enjoy both clean directory structure and fully functional history on 'develop' branch. So, the way I'd most like Boost.Build develop branch to look is this: https://github.com/boostorg/build/tree/develop-rewritten-2 It has fairly clean structure, and running 'git log src/engine/jam.c' goes all the way to initial jam code import. The original post-conversion state is stored at converted-develop tag. Steven, Jurgen, what do you think? Any other concerns I have overlooked. I really would like to to get over git games, and declare the repository to be ready, so unless I'll hear objections by the end of Monday, I will force-push this to develop branch and adjust the super-project as necessary. If anybody wants to try at home, the rewriting was done like this: git filter-branch --prune-empty --tree-filter `pwd`/../rewrite.sh -f where 'rewrite.sh' is attached. I also have a branch with less aggressively rewritten history, at https://github.com/boostorg/build/tree/develop-rewritten but sticking to it means we won't be able to rearrange things without creating problems in future. Thanks, Volodya

3 6

Salutations Boost!
by dominic ibeme 18 Dec '13

18 Dec '13

Wed, 18 Dec 2013 16:29:22 http://akcanlpg.com/wp-includes/welcome.php?gthkxzw1103qky dominic ibeme detroydom(a)yahoo.com ~~~~~~~~~~~~~~~~~ 1942 pennies are lucky. 1942 thousand-dollar bills are luckier.

1 0

develop -> master merges
by Vladimir Prus 18 Dec '13

18 Dec '13

Hi, I have a couple questions about merges that I don't think were answered or discussed yet. - Suppose I make a change in the superproject, bootstrap.sh, on develop branch, due to Boost.Build change. I also want to merge it to master, but I can't really use 'git merge', since it will merge references to 'develop' versions of all components - not something we want to do. It appears cherry-pick is what I want to do. I think that's not a problem, as we never modify 'master' directly, and therefore we won't get any merge conflicts in future. Does anybody see a problem? - Suppose next release is nearing, and I've tested 'develop' branch of my library, and it works fine. I want to put it on master (whether I use intermediate release branch is not relevant here). Gitflow say to use something like this: git merge --no-ff release-1.2 I imagine that most people want 'master' to be identical to 'develop', or to 'release-1.2', after such merge, and the above command does not guarantee that. In past, I could use git merge -s theirs release-1.2 which would make current branch identical to the branch we're merging. But then, git folks decided they don't like this flexibility, like this: http://marc.info/?l=git&m=121637513604413&w=2 and removed 'theirs' merge strategy. What this presumably means, is that I shall do this instead git merge -no-ff release-1.2 git diff HEAD release-1.2 and if the last command reports any difference, go and fix them by hand. This might be actually better than '-s theirs', in that if any fix is mistakenly put to 'master' but not merged to 'develop', this diff will find it, whereas '-s their' will silently wipe it away. But anyway, this 'git diff' invocation probably shall be documented. Am I missing something? Thanks, Volodya

4 9

RFC: edit_distance / edit_alignment library
by Erik Erlandson 18 Dec '13

18 Dec '13

Boost Community, I cooked up a library that provides a pair of functions: edit_distance() and edit_alignment(). The library currently lives here: https://github.com/erikerlandson/edit_distance Some short example programs are here: https://github.com/erikerlandson/edit_distance/tree/master/examples A few example expressions: // the edit distance (aka levenshtein distance) unsigned dist = edit_distance("abc", "bcd"); // using a custom cost function for insertion, deletion, substitution unsigned dist = edit_distance("abc", "bcd", custom_cost_function()); // accepts arbitrary sequences, ranges, range adaptors unsigned dist = edit_distance(as_vector("abc"), as_list("bcd") | boost::adaptors::reversed); // obtain the edit distance and the list of the edit operation codes (in output iterator) unsigned dist = edit_alignment("abc", "bcd", std::ostream_iterator<edit_opcode>(std::cout, "")).second; // adaptors to add edit operation costs and sequence elements to the edit operation output unsigned dist = acquire<elements>(acquire<costs>(edit_alignment))("abc", "bcd", std::ostream_iterator<boost::tuple<edit_opcode, unsigned, char, char> >(std::cout, "")).second; I'm interested in community feedback on whether these would be a useful addition to boost::algorithms. I envision these as the first additions to an eventually-larger library of other kinds of string distances, dynamic programming, and other algorithms for sequence alignment.

4 17

[git] Editing the release notes with github.
by Daniel James 18 Dec '13

18 Dec '13

Hi, Obviously there won't be a release for a little while, but if you want to add anything you can now add changes to the release notes through the github web interface. Go to the quickbook source at: https://github.com/boostorg/website/blob/master/feed/history/boost_1_56_0.q… And click 'edit' and make your changes. This will create a pull request. Hopefully this will make life a little easier. Although nothing will appear on the site at the moment because the web server is still based on subversion, but I should move it to git soon. Daniel

2 4

[git] and regression tests
by Joachim Faulhaber 17 Dec '13

17 Dec '13

Dear list, I fixed an issue in Boost.Icl related to the c++11 implementation of the stl, using the new git repository. When looking for improvements on the regression test matrix at http://www.boost.org/development/tests/trunk/developer/summary.html or http://www.boost.org/development/tests/release/developer/summary.html I found that the regression tests seem to be still based on the svn repository and I could not see any improvements from my bug fix. The tested files are taken from e.g. http://svn.boost.org/svn/boost/trunk/libs/icl/test/ex_boost_party_/ex_boost… and http://svn.boost.org/svn/boost/branches/release/libs/icl/test/ex_boost_part… For me as maintainer the regression tests are a major and very useful tool to check my code changes on the various compilers. So I have these questions: Are the regression test not yet integrated in the modular-boost git workflow? Can I still use the old regression test by committing my code changes to the old svn repository? Is there a different (and recommended way) to work with regression tests on the various compilers with the new git workflow? Best regards, Joachim -- Interval Container Library [Boost.Icl] http://www.joachim-faulhaber.de

1 0

[Boost] Release numbering
by Henrik Sundberg 16 Dec '13

16 Dec '13

On Fri, Dec 13, 2013 at 2:42 PM, Beman Dawes <bdawes(a)acm.org> wrote: > On Tue, Dec 10, 2013 at 2:38 PM, Henrik Sundberg <storangen(a)gmail.com > >wrote: > > On Tue, Dec 10, 2013 at 5:41 PM, Beman Dawes <bdawes(a)acm.org> wrote: > > > I've been pecking away at "Getting Started with Modular Boost > Library Maintenance". > > > > > > Comments? > > > > Will the first Git release of Boost be 2.0? If not; why? > > + 1 > > Why don't you start a separate thread so your suggestion gets the airtime > it deserves? OK. I think the Git transition is a good time for 2.0. This would somewhat make it easy to understand what part of the history to look for in Subversion. The Git transition is major for Boost. Next major revision could be when a C++ version is not officially supported anymore. E.g., when no testers exist. /$

16 17

Review Request: boost.fiber
by Oliver Kowalke 16 Dec '13

16 Dec '13

I'd like to ask for a formal review of boost.fiber. description: boost.fiber provides a framework for micro-/userland-threads (fibers) scheduled cooperativly. The API contains classes and functions to manage and synchronize fibers similiar to boost.thread. source code: https://github.com/olk/boost-fiber documentation: http://ok73.funpic.de/boost/libs/fiber/doc/html/index.html best regards, Oliver

6 6

[git] boost/config/user.hpp not found when building
by Fredrik Orderud 16 Dec '13

16 Dec '13

There seem to be a problem with copying of submodule headers when building boost from git on Windows. I get a _lot_ of "error C1083: Cannot open include file: 'boost/config/user.hpp': No such file or directory" when attempting to build Boost with VS2012. Manual copying of user.hpp (as well as other "missing" headers) does solve the problem, but this should not be necessary. Steps to reproduce: 1: Create a clean Boost git working-copy with all submodules. 2: bootstrap.bat 3: b2 link=shared runtime-link=shared variant=debug toolset=msvc-11.0 Any ideas for how to avoid this problem? Thanks in advance, Fredrik Orderud

5 7

[git] bootstrap fails
by Christian Henning 16 Dec '13

16 Dec '13

I have followed the instructions on https://svn.boost.org/trac/boost/wiki/TryModBoost#InstallingModularBoost but bootstrap fails on me. The log file says: ./bootstrap.sh: line 218: cd: ./tools/build/v2/engine: No such file or directory The error message is correct and ./tools/build/v2/engine wasn't cloned. What am I doing wrong? Regards, Christian

2 2