0
I Use This!
Inactive

Commits : Listings

Analyzed 2 days ago. based on code collected 2 days ago.
Aug 19, 2024 — Aug 19, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Add support for column store with variable-sized chunks on the leaf pages. Fix bugs in the hazard pointer implementation, it assumed WT alway released pages in LIFO order, which wasn't correct. Move hazard pointer support into the cache layer -- most of it is already there anyway. More... over 15 years ago
Add support for column store with variable-sized chunks on the leaf pages. Verify don't reference past the end-of-file before trying to read the pages. It's a simple check and gives better diagnostics. More... over 15 years ago
Move the debugging function to dump the ddescription table into the debug file. More... over 15 years ago
Add support for column store with variable-sized chunks on the leaf pages. Rename debug-dump functions from dump_XXX to debug_XXX. Pass a WT_TOC to debug-dump routines (assumes debugging on pages happens when using a WT_TOC, which is generally true, and brings these routines into line with other APIs). Avoid creating a WT_TOC repeatedly when dumping a tree, create it once and pass it around. More... over 15 years ago
Minor comment change. More... over 15 years ago
Use message output routines instead of printf. More... over 15 years ago
rework the cache dump code to put each page on its own line, with addr & size information. More... over 15 years ago
Fix a comment. More... over 15 years ago
Minor renaming of the "open verify" routine, it's not just about sizes. More... over 15 years ago
Lint More... over 15 years ago
Typo, swapped the DB/IDB handle & database statistics. More... over 15 years ago
Minor fix to bucket count. More... over 15 years ago
Add verify routines for a few more setter methods. More... over 15 years ago
Rename env/{engine,toc}.c to env/env_{workq,toc}.c for consistency. More... over 15 years ago
Move statistics strings into their own class file in dist so we can add config information. Specifically, stat fields can now be marked "perm", which means they aren't cleared by the stat clear method. This means cache allocation information can be moved into the stat structures and displayed normally. This also means statistics are required in the system (although it would probably be pretty easy to only maintain "perm" stat fields instead of all stat fields if we ever need a smaller footprint). More... over 15 years ago
remove whitespace from s_XXX scripts. More... over 15 years ago
whitespace More... over 15 years ago
Remove trailing whitespace from python scripts. More... over 15 years ago
Add tuning method to set the size of the cache's hash bucket array. Regularize the naming of size getter/setter methods -- it's XXX_size_get, not XXX_max_get. More... over 15 years ago
Add tuning method to set the size of the cache's hash bucket array. Regularize the naming of size getter/setter methods -- it's XXX_size_get, not XXX_max_get. More... over 15 years ago
Lint. More... over 15 years ago
Switch lockless programming style from epoch to hazard references. Use an array of pointers for the WT_TOC list. Add method to set/get the number of WT_TOC structures supported. Add the ability to explicitly acquire an IENV mutex in the API layer. More... over 15 years ago
Add support for statistics at a finer granularity. More... over 15 years ago
Switch lockless programming style from epoch to hazard references. More... over 15 years ago
Switch lockless programming style from epoch to hazard references. Add the ability to explicitly acquire an IENV mutex in the API layer. Add method to set/get the number of WT_TOC structures supported. More... over 15 years ago
Add the ability to explicitly acquire an IENV mutex in the API layer. More... over 15 years ago
Make it possible to turn Huffman encoding off. Use putenv (which is portable) instead of _malloc_options, which isn't, to set the FreeBSD malloc options. More... over 15 years ago
We now have a real env, errors have been fixed to use it. More... over 15 years ago
Last version with cache algorithm where threads wait for the cache thread and we use the lowest API generation to decide when they can proceed. In other words, epochs -- bug remains where the linked list references aren't volatile, so I keep finding threads accessing freed pages. I'm going to switch over to hazard pointers, hopefully. More... over 15 years ago
Fix a memory leak in the Huffman code. Change __wt_free to atomically clear the reference location, then overwrite memory in a cleaner fashion, no macro needed. More... over 15 years ago