Fossil

Timeline
Login

Timeline

Many hyperlinks are disabled.
Use anonymous login to enable hyperlinks.

50 events by user aku occurring around 2007-10-06 03:53:41.

2007-10-14
01:58
Reworked the whole handling of meta data (author, commit message, plus project/branch information), so that revisions now store only the meta id, everything else is stored centrally. All the relevant pieces (author, cmessage, symbols, projects) now also get numeric ids assigned early instead of when being saved to the state. Project ids are loaded from the state now too. ... (check-in: 67c24820c7 user: aku tags: trunk)
2007-10-13
23:57
Completed loading of pass I from persistent state when skipped. ... (check-in: 54ac684df7 user: aku tags: trunk)
23:42
For the record an alternative design for the main method of the pass manager where the four principal loops (skip, setup, run, and defer) are merged into two (skip/setup) and (run/defer). Not really less code, also more complex to understand and maintain. The fact that 'setup' and 'run' is performed for the same passes is more difficult to see. This code will not be used. ... (Closed-Leaf check-in: 4df56a9e7e user: aku tags: trunk)
23:29
Extended pass manager to handle the skipped and defered passes coming before and after the actually executed passes. Extended passes I and II to have the required methods. Implemented loading (for skipped passes) as skeletons, implemented discarding (for defered passes) completely. Extended state manager with ability to discard state. ... (check-in: ae19c0fcb4 user: aku tags: trunk)
21:15
Extended pass I to capture the 'file executable' info of rcs archives. Currently the only way to store this info in the destination will be the use of fossil tags. ... (check-in: 27470a9304 user: aku tags: trunk)
05:39
The basic work of setting up and linking revisions, tags, and branches is complete. ... (check-in: cb70cf4ad6 user: aku tags: trunk)
2007-10-12
07:20
Extended rcs parser with another callback run after the processing of the revision tree completed, before starting on description and delta texts. Updated file class to intercept the callback. ... (check-in: 40eaf58ec7 user: aku tags: trunk)
07:18
Continued to flesh out revision and symbol processing. Started on the project level classses. ... (check-in: da9295c6f6 user: aku tags: trunk)
05:57
Extended rcs parser to tag errors which are syntax errors in the RCS archive. Extended pass handling to capture trouble and distinguish between internal problems and syntax errors. ... (check-in: be891232a2 user: aku tags: trunk)
05:56
Started to flesh out the handling symbols and revisions coming in from the rcs parser. First fragments of revision object. Type methods for classification of revision numb ers. ... (check-in: bd131addb9 user: aku tags: trunk)
2007-10-10
05:36
Added more checks to pass I looking for possible file/directory conflicts in the repository, and modified traverser to report on and ignore sub directories of Attic's. ... (check-in: a10f654ac7 user: aku tags: trunk)
04:23
API tweak rcs parser, definition of symbols, call through the callback once per symbol instead of once for the whole set. ... (check-in: 338854a61a user: aku tags: trunk)
03:28
Added a lot of skeleton files for the revision and symbol data structures, for both project and file level. ... (check-in: 84de38d73f user: aku tags: trunk)
2007-10-06
22:13
Found easier way to access the typevariables of singletons when writing the rcs parser. Now moved the other singletons to the same construction. ... (check-in: 70b0aa899a user: aku tags: trunk)
22:06
Broke static dependency cycle repository <- project <- file <- repository, now using dynamic/runtime object references instead. This allows us to keep the nice and readable format for our files, code is readable too. ... (check-in: 8487172254 user: aku tags: trunk)
21:59
Started capture of revision information in file objects. Capturing authors and commit messages and repository level. Completed persistence for these latter too. Rearranged the requirements, imports, and exports a bit to handle the new dependency cycle repository <- project <- file <- repository ... (check-in: 3d88cfd05d user: aku tags: trunk)
21:19
Continued work on pass II, skeleton of it completed by adding outline of code for persistence. ... (check-in: 6d1811d61e user: aku tags: trunk)
21:01
Added skeleton and dump methods to make file objects proper sinks for the rcs parser. ... (check-in: 2e3815c3b4 user: aku tags: trunk)
21:00
rcs parser rewritten to snit type, singleton, using callbacks to convey information. Got a grip on regexp -start -indices (^ -> \A, indices are absolute). This means that I now don't have to shift the unprocessed string down in memory, it is all index processing now. This makes for a much faster parser, especially for large files with many revisions (example: ChangeLogs). ... (check-in: e7bb3d073d user: aku tags: trunk)
20:58
Nicer parser integration into pass II, and sort files for easier filesystem traversal during parsing. ... (check-in: ec053168a8 user: aku tags: trunk)
18:43
Fleshed out pass II, added skeleton of rcs archive class, started integration of rcs parser. ... (check-in: 8a93ffa9c1 user: aku tags: trunk)
18:28
Bringing back in the old parser for rcs archives. ... (check-in: 860f533e98 user: aku tags: trunk)
03:56
Added pass II (collection of revision and symbols), basic skeleton and state definitions, and integrated into the pass manager. ... (check-in: fdbc01df95 user: aku tags: trunk)
03:55
Added documentation to state definition, and debugging output to the state system itself. ... (check-in: 10f9d51bb2 user: aku tags: trunk)
03:53
Better formatting for --help-passes, align codes. ... (check-in: 43d72c6246 user: aku tags: trunk)
2007-10-05
07:27
Completed integration of state with pass I, now saving projects and files into it. ... (check-in: 042d54bae5 user: aku tags: trunk)
07:08
Collapse multiple uses of a project path into one project. ... (check-in: d174affb48 user: aku tags: trunk)
07:07
Added documentation for the persistent state of pass I. ... (check-in: 78da61db0e user: aku tags: trunk)
06:50
Extended state handling with methods to declare usage and structure of state, started integration of state with pass I, collection of projects and files. ... (check-in: fb1e36d290 user: aku tags: trunk)
05:33
Added the basic parts of the state manager and integrated it with option processor and pass manager. ... (check-in: eb656de7d9 user: aku tags: trunk)
2007-10-04
04:34
Continued work on pass I. Filled in the repository management, and basic implementation of project objects. Missing are persistence and the foundation for that (cache database). ... (check-in: 52f2254007 user: aku tags: trunk)
04:32
Extended option processing, added handling of verbosity/quietness. ... (check-in: 2929a438ef user: aku tags: trunk)
2007-10-03
04:18
Added stub for repository management, added most of pass I, collecting archives in the repository. ... (check-in: 47740cc1f6 user: aku tags: trunk)
2007-10-02
06:48
Added the pass management, integrated with application and option processor. ... (check-in: 5911515322 user: aku tags: trunk)
05:33
Re-added the user feedback and error reporting utilities, with modifications, and completed the handling of the informational options. ... (check-in: d57b7b4a05 user: aku tags: trunk)
03:05
Third attempt at getting a cvs importer which can handle branches. Using cvs2svn code and design notes as a guide. ... (check-in: 2a98ac44bd user: aku tags: trunk)
2007-09-27
04:44
CVS import. First, fixed sig::next regarding two things. One, we have to take the root version of a file into account as a possible predecessor. Two, a missing changed file may be misclassified and actually be added instead. Second, modified the search for a root changeset of a branch. We now try the existing regular intersection first for exactness, and in case of failure we fall back to a voting scheme to locate the most acceptable aka non-conflicting changeset. ... (check-in: 7a64b9e738 user: aku tags: trunk)
2007-09-26
07:06
Bugfix for 3852590ce6. Make the use of cache files in the rcs parser actually dependent on the user option. ... (check-in: f6f5c902e6 user: aku tags: trunk)
05:06
Continued work on the import of branches. Main principle now is to handle the branches vertically. First the trunk, then the branch starting with the first unprocessed changeset, and so forth. Looks more promising than the previous approach. Currently handles just a bit over half of the test projects (11 of 21 in 6 repositories). ... (check-in: 72dac950c3 user: aku tags: trunk)
05:02
New feature for importer. rcs parser extended so that it can store parse results for quick loading in future runs. This feature has no real use in regular use of the importer, i.e. one-shot conversion of a CVS repository to fossil. It is however useful for debugging when the source repository is scanned many times during test runs. Especially for large files, with lots of changes (like ChangeLogs), the direct loading of a Tcl dictionary is much faster than actually parsing the archive files. ... (check-in: 3852590ce6 user: aku tags: trunk)
2007-09-20
07:37
Entered the general structure planned for processing of branches. Incomplete. This code right now exits when it finds branch csets. Some debug output to see detailed internals from which to pull the pieces together. ... (check-in: 25bc721076 user: aku tags: trunk)
07:33
Trim the commit messages to remove irrelevant leading and trailing line-endings. ... (check-in: a9dcf09124 user: aku tags: trunk)
07:14
Got rid of the explicit revision tree and rephrased the trunk processing to use a loop which is more self-explanatory. Started to add in code needed when we process the branches as well, currently they will have now effect. ... (check-in: cbbf9a7575 user: aku tags: trunk)
07:08
Moved the counting of imported changesets to the beginning for a more accurate percent calculation and proper ending at 100%. ... (check-in: 330f2da791 user: aku tags: trunk)
03:51
Modified sorting of timeline entries for the same second to properly split files and file versions from each other, and to have newer revisions later. Further added a storage for error messages to be repeated when the importer exist. First user is the code reporting corrupted archive files detected during a checkout. ... (check-in: a5476aed27 user: aku tags: trunk)
2007-09-17
07:05
Further work on the CVS frontend. The main parts for doing the extraction and management of changesets are now in a separate package. ... (check-in: ae54e928c2 user: aku tags: trunk)
04:14
Fixed bug in the new code setting up the timeline, forgot to clean up loop state, causing data from previous iterations to bleed into the current one, causing the use of bogus data and ultimatively a bogus timeline. Note! Incidentially a bug in the old code was fixed by the new one, the use of mismatching revisions and commit messages. That caused the generation of extra changesets in the old code. ... (check-in: 10e3b3ed76 user: aku tags: trunk)
03:13
Cleaned up the creation of the workspace directory. ... (check-in: 2740b48b63 user: aku tags: trunk)
03:03
More structuring of the CVS backend, encapsulated the management of the global timeline of events in the project in a separate package. ... (check-in: 1593006ef3 user: aku tags: trunk)
01:43
Added structure to the CVS frontend code, putting the repository traversal into its own package. ... (check-in: 6f121db1e2 user: aku tags: trunk)