Many hyperlinks are disabled.
Use anonymous login
to enable hyperlinks.
50 most recent check-ins
|
2025-06-07
| ||
| 00:50 | Add a missing stmt finalize in test code which, incidentally, triggers a memory corruption panic in Tcl (but certainly my fault). Leaf check-in: 0e270edf3e user: stephan tags: tcl-stmt-cmd | |
|
2025-06-06
| ||
| 22:56 | Doc tweaks. check-in: 91d1478636 user: stephan tags: tcl-stmt-cmd | |
| 22:00 | Internal cleanups and docs. check-in: df4155da25 user: stephan tags: tcl-stmt-cmd | |
| 21:39 | Add the (db.stmt get -list/-dict) flags. check-in: e5e4da0305 user: stephan tags: tcl-stmt-cmd | |
| 21:07 | More work on the prepared stmt command objects. Started documenting them. check-in: 3b4967ab92 user: stephan tags: tcl-stmt-cmd | |
| 18:30 | Eliminate the discrepancy mentioned in the previous checkin: closing a db object invalidates stmt objects it prepared rather than keeping the db open via refcounting. This keeps the underlying tcl-binding object's alive, because we can't currently manage the QueryState lifetimes without that, but their db connection is closed and the resources on that side of the API are freed (the rest is freed via Tcl's normal lifetime management). check-in: b2fc76cc2d user: stephan tags: tcl-stmt-cmd | |
| 17:35 | When closing a fsl_cx, invalidate any opened query command objects so that they do not step in invalid memory later on. They will trigger an error when used for anything except their finalize method. fsl_db contexts, on the other hand, share a refcount and will stay open until the last query is closed. This discrepancy is unsightly, but the problem child is the former case - the fsl_cx internally manages its db handles and we can't influence them from here like we can the db handles. Because fossil cx's can be re-used after closing them, we're bound to honor their close method and actually close the SCM dbs, which invalidates opened statements. For fsl_db contexts, though, we can(?) justify keeping the db handle open so long as there's a reference to it. Maybe. Maybe not. check-in: 820c64211e user: stephan tags: tcl-stmt-cmd | |
| 15:34 | tcl: add (incomplete) support for prepared statements as command objects. What's been tested (which isn't everything) seems to work but there is still a lifetime quirk involving prepared statements to fossil-managed db handles when the fossil instance is closed before the statement (which shouldn't happen, and doesn't for non-fossil db handles, but we need a small extra piece of infrastructure to account for this case). check-in: b4c9938909 user: stephan tags: tcl-stmt-cmd | |
| 09:07 | Rename the redunantly-named tclfossil.tcl to fossil.tcl. Leaf check-in: b4b27f57df user: stephan tags: trunk | |
|
2025-06-05
| ||
| 14:27 | Remove some unused code. check-in: 3818a14ac6 user: stephan tags: trunk | |
| 10:03 | Internal cleanups in the tcl pieces. check-in: e14034f9ea user: stephan tags: trunk | |
|
2025-06-04
| ||
| 21:47 | Tcl doc fix. check-in: e7b34d933b user: stephan tags: trunk | |
| 09:35 | Re-do the (fossil info) command with a different interface. check-in: ee4e191a1f user: stephan tags: trunk | |
|
2025-06-03
| ||
| 20:35 | Tweak the tcl test script to be able to run individual external test scripts, rather than the whole suite. Move most of the tests into external files. check-in: d1709bd835 user: stephan tags: trunk | |
| 18:57 | Minor internal cleanups. check-in: 125aec9b39 user: stephan tags: trunk | |
| 18:37 | Refactor th_cmdo to have a refcount so that we can eventually do things like create prepared queries as command objects and keep their lifetimes kosher vis a vis their databases' command objects. Reminder to self: a query command object would provide a subset of db operations relevant to the query, e.g. binding, stepping, fetching result columns, and finalizing the underlying prepared statement (which would also decrement the associated db's refcount). check-in: 1d2131fb80 user: stephan tags: trunk | |
| 15:07 | tcl: add -size, -renames, -time, and -localtime flags to the (fossil ls) command. check-in: 84d048fb85 user: stephan tags: trunk | |
| 15:07 | Add fsl_content_size_v2(). check-in: 26b7c660a0 user: stephan tags: trunk | |
| 13:58 | No, really this time. check-in: 1ec5e365a0 user: stephan tags: trunk | |
| 13:54 | Account for the renamed tcl extension in shell.c. check-in: 8a86427869 user: stephan tags: trunk | |
| 12:32 | Rename the tcl extension from tclfossil to simply libfossil. check-in: 3fd5067559 user: stephan tags: trunk | |
| 12:20 | The fix mentioned in the previous checkin. check-in: cb95eda67a user: stephan tags: trunk | |
| 12:19 | Move the tclsh interactive shell code into its own file. This will unavoidable break a test which counts files from the repo, but that fix will be in the next checkin. check-in: 626a32ed4c user: stephan tags: trunk | |
| 11:44 | Move the (fossil ls) tcl command's code into its own file. check-in: 4ea91b8b37 user: stephan tags: trunk | |
| 11:42 | Fix a cosmetic spacing issue in the generated makefile rules and move the relevant tcl code from auto.def to wh-common.tcl for reuse by the tcl binding's configure script. check-in: 8420e1a1e3 user: stephan tags: trunk | |
|
2025-06-02
| ||
| 16:15 | Tcl API renaming and internals cleanups. check-in: 44d20f7706 user: stephan tags: trunk | |
| 11:52 | Add a placeholder stub for the (fossil file) tcl command. check-in: 943fe9dd62 user: stephan tags: trunk | |
| 10:33 | A long-overdue s/th_flags_parse_t/th_fp/g. check-in: d28f8a1041 user: stephan tags: trunk | |
| 10:03 | Add the beginnings of a (fossil checkout) command, the only current subcommand being scan, which rescans the checkout for local changes. check-in: 678ba2d255 user: stephan tags: trunk | |
| 10:02 | Replace some switch fall-through comments with FSL_SWITCH_FALL_THROUGH. check-in: 5b2128d304 user: stephan tags: trunk | |
| 09:09 | Random doc touchups. check-in: f11f20e8e3 user: stephan tags: trunk | |
| 07:55 | tcl: rename :ckout: to :checkout: because if we don't then we'll eventually feel compelled use ckout consistently everywhere as the One True Name of checkout-related subcommands and flags. Cosmetic tweaks in tcl test output. check-in: c16edbf03a user: stephan tags: trunk | |
|
2025-06-01
| ||
| 21:08 | Internal docs and increase the default command flag limit for the ftcl_fp_... macros because the ls command was right on the border of the old value. check-in: e1d1c52e6f user: stephan tags: trunk | |
| 20:52 | Remove some stray debug output. check-in: 9ad9ce812f user: stephan tags: trunk | |
| 20:45 | Remove the var(fossil) entry from the array set up for (fossil ls --eval), as the eval has direct access to the calling object via its local var name (if any). check-in: a4d59498cd user: stephan tags: trunk | |
| 20:23 | Add (fossil ls ---eval varName script) flag and start documenting the (fossil ls) subcommand. check-in: 1543a5ac69 user: stephan tags: trunk | |
| 16:51 | Cache more array keys in (fossil ls). check-in: 0fb2c050dc user: stephan tags: trunk | |
| 15:45 | More work on the (fossil ls) command. check-in: e89b50e785 user: stephan tags: trunk | |
| 14:39 | Add the beginnings of a tcl (fossil ls) command. check-in: aa56e240b8 user: stephan tags: trunk | |
| 14:38 | Minor API doc updates. check-in: 16ab025029 user: stephan tags: trunk | |
| 11:06 | Add tcl (db query --null-string X) to set the query's string representation of SQL NULL to X. Tinkering with a new set of macros for command flag parsing. check-in: c185c5b613 user: stephan tags: trunk | |
|
2025-05-31
| ||
| 13:29 | Document the tcl (db config) subcommand and improve its return value semantics a bit. check-in: 763f206348 user: stephan tags: trunk | |
| 13:00 | Add tcl db config --null-string X and -null-string (getter) flags to get/set the db's current string representation of SQL NULL. check-in: 1451c750b1 user: stephan tags: trunk | |
| 10:49 | Remove the -Wno-unused-parameters build flag in --pedantic mode, formerly needed for sqlite3.c, as that's been resolved upstream. check-in: 141248ce0f user: stephan tags: trunk | |
| 10:48 | Correct the RID on the parent: line of f-status output. check-in: f6827caaf1 user: stephan tags: trunk | |
| 09:44 | Pull in latest sqlite3.c to get it compiling with -Wextra -pedantic. check-in: 9b35d809f8 user: stephan tags: trunk | |
| 09:33 | Re-add -Wextra -pedantic CC flags, and work out its complaints across many files, but they're disabled without the --pedantic flag because they break with the upstream fnc.c and sqlite3.c. check-in: 84ef1197c3 user: stephan tags: trunk | |
| 09:32 | Speed up the lambda-to-name conversion for the common cases. check-in: c3c3c7767b user: stephan tags: trunk | |
| 09:31 | Pull in upstream sqlite3.c/h. check-in: 3ec8fdf670 user: stephan tags: trunk | |
|
2025-05-30
| ||
| 16:59 | Simplify proj-parse-flags a bit. check-in: da40a61cb6 user: stephan tags: trunk | |