Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
11 check-ins related to "annotate_noleak"
2011-10-14
| ||
16:11 |
Merging the annotate_noleak changes, about removing an important memory leak in
the annotate operation.
It also fixes some blob behaviour in blob.c and content.c. Update: Removed from trunk. Replaced by the must simpler fix at [9929bab702f99839ee] ... (check-in: 409f370a user: viriketo tags: declined) | |
2011-09-13
| ||
17:18 | The Label list was not properly freed in some cases; looking at the annotate of 'win/Makefile.msc' was triggering the case. ... (Closed-Leaf check-in: 4860d9f2 user: viriketo tags: annotate_noleak) | |
16:43 | Getting in the changes reducing the leak on annotate. In the fossil mailing list, I received only opinions in favour of getting mallocs and frees properly paired, so I think noone will refuse the change. ... (check-in: 7870a89b user: viriketo tags: annotate_noleak) | |
2011-09-06
| ||
20:30 | Documenting the Label structure I introduced. ... (Closed-Leaf check-in: 58a02a2e user: viriketo tags: label_linkedlist) | |
2011-09-04
| ||
22:14 |
Implementation of a linked list to solve the memory leak described in a TODO
in [e2ebb1f5cae8].
This code is slower than having the memory leak, and at the end, it was not a big memory leak. Let's say, 10 byte per revision involved in a file annotate. If a file has 30000 revisions, it may go around 300KB then. For this leak to be noticeable (as we have a content cache until 50MB), it would mean having above a milion revisions for the file annotated. So, I keep this in a branch apart, as maybe the leak is worth the faster code. I have not measured the speed difference though. ... (check-in: ef8266b7 user: viriketo tags: label_linkedlist) | |
21:39 |
Reducing the leak of annotate even more.
I wrote a TODO in the file, for a leak still present and relevant for files modified by many versions, but that requires harder work to get right. ... (check-in: e2ebb1f5 user: viriketo tags: annotate_noleak) | |
21:02 | Making annotate not segfault for the complex test case I have. There is still a little leak... but muuuch smaller than it was. ... (check-in: fa760a24 user: viriketo tags: annotate_noleak) | |
20:43 | Updating from trunk to get the latest code. ... (check-in: 73b6ff49 user: viriketo tags: annotate_noleak) | |
20:41 |
What I did to get the 'annotate' command not leak.
Notice that I disabled again the check for blob_is_reset, as in trunk. ... (check-in: 7ad4dfbe user: viriketo tags: annotate_noleak) | |
01:28 | Update the built-in SQLite to the latest 3.7.8-alpha version that contains the improved merge-sort logic. ... (check-in: 0cf54160 user: drh tags: trunk) | |
2011-09-01
| ||
22:22 | Fix typos and other cleanup on the "build.wiki" doc page. ... (check-in: 0f9fb203 user: drh tags: trunk) | |