summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2007-05-08add per-action state accessor apiHEADasyncRay Strode2-14/+97
We want to keep a solid distinction between the three types of data that are used in a transaction. One is per-transaction, inter-action state. This is currently handled by gobject data on the transaction object. Another is per-action input arguments. These come from the initiator of the transaction and are passed to the action every time the action is run. The last is per-action internal state. Since action processing and rollback functions are reentrant (they get called until the action says its complete) each action needs to some how track what work its already done and what work it needs to do to be finished. Previously, this was handled ad-hoc as per-transaction gobject data, but now there is a dedicated api.
2007-05-07i don't knowRay Strode1-20/+54
2007-05-07allow pop_transaction_wait/resume to get called beforeRay Strode1-12/+3
transaction is commited
2007-05-07random fix ups in the test case to make it buildRay Strode1-4/+5
2007-05-07Add missing variable declarationRay Strode1-0/+2
2007-05-06rework the test case to have two separate transactions,Ray Strode1-14/+98
an input transaction that reads from stdin and an output transaction that writes to stdout. Next up is making them cooridinate with each other, so the write transaction writes after the read transaction fills up its buffer or blocks waiting for more data.
2007-05-06add some comments explaining two error cases in theRay Strode1-0/+6
io test action
2007-05-06add new api to set the transaction error based on errnoRay Strode2-0/+13
2007-05-06repeat a node if child nodes get added to itRay Strode1-32/+52
use common buffer for read and write action in test case
2007-05-06insert actions as dependencies of the root node,Ray Strode1-1/+8
if there isn't a current node (i.e., before the user does commit)
2007-05-06store actions as a tree internally, to have betterRay Strode1-53/+260
accounting of the dependencies between action. Not tested yet, so probably doesn't work.
2007-05-06change test case from file copy to catRay Strode1-159/+61
2007-05-06use raw encoding on io channel so buffering can be disabledRay Strode1-0/+1
2007-05-06move some functions around so that public functionsRay Strode1-197/+197
are at the bottom.
2007-05-06add docs to propertiesRay Strode1-1/+20
2007-05-06move things around a bitRay Strode1-82/+74
2007-05-06comment the transaction private structure betterRay Strode1-5/+45
2007-05-06don't assume pop_transaction_resume is unpausing a transaction paused with ↵Ray Strode1-3/+5
wait_a_while or wait_for_fd
2007-05-06Make it build againRay Strode1-1/+1
s/status/action->rollback_status/
2007-05-06add prototypes for all static functionsRay Strode1-3/+31
2007-05-06save action status on action object. If we ever come back to an action that'sRay Strode1-87/+175
already finished (because say it added some actions to the queue before finishing and those actions finished), then just return. Drop lame test case and start to write one that is more real-world (will eventually copy a file)
2007-05-06Fix up formattingRay Strode7-721/+1021
Run source code through "uncrustify"
2007-05-05add new untested read and write actions to test case area.Ray Strode2-2/+195
Also add new result and error properties on the transaction object
2007-05-03add untested support for transactions within transactions andRay Strode2-103/+279
a pile of new public api to be able to support it
2007-05-03add random thought about dependent actionsRay Strode1-1/+8
2007-05-03allow one action to depend on anotherRay Strode3-11/+53
add pop-transaction to build system random fixups
2007-05-03add a few more thoughts to the design docRay Strode1-9/+9
2007-05-03add assertsRay Strode2-4/+45
2007-05-03misc fixesRay Strode2-43/+41
drop unused Error object clean up test case track when we are calling back to the user better
2007-05-02More cleanups, make it workRay Strode1-74/+126
drop wait for source, it can't work because event sources don't have a generic hook onto dispatch api and callback signatures vary from event source to event source add lame test case
2007-05-02just chugging along on the transaction stuff...Ray Strode2-151/+275
2007-05-02flesh out the transaction bits a little moreRay Strode1-13/+120
2007-05-02flesh things out a bit more in the transaction implementation (still not ↵Ray Strode3-56/+352
done or even compilable)
2007-04-30flesh out KEEPING-PROGRAMS-LINEAR a little more andRay Strode3-3/+381
create stub objects to implement the ideas
2007-04-30fix a few typos and add some additional api ideasRay Strode1-14/+31
2007-04-30add braindump on how to keep programs sequentialRay Strode1-0/+173
but still asynchronous
2007-04-30change readiness to ready in function nameRay Strode1-11/+7
2007-04-18port more things over to use the new async error handlingRay Strode5-35/+98
2007-04-18Start to move code over to PopXReplyWatch instead ofRay Strode4-57/+176
gdk_error_trap_push/gdk_error_trap_pop
2007-04-18add experiment x reply event sourceRay Strode2-0/+359
2007-04-08add a little blurb about error handlingRay Strode2-1/+39
2007-04-08Merge branch 'event-rewrite'masterRay Strode30-1108/+4306
Conflicts: pop-window.c
2007-04-08put a placeholder view up before the window gets its first damage eventRay Strode2-4/+73
2007-04-08Try to draw lessRay Strode6-29/+229
Make some motions toward implementing what is described in FIGURING-OUT-WHAT-TO-REDRAW
2007-04-07Drop un-needed function declarationRay Strode1-1/+0
The PopWindowView class had a method pop_window_view_queue_damage declared in its header that was never used or defined.
2007-04-07add config.h to the top of every source fileRay Strode7-6/+8
2007-04-07random thoughts on how to draw less each time aroundRay Strode1-0/+41
2007-04-06random, somewhat spurious cleanup.Ray Strode3-29/+17
2007-04-02i dunnoRay Strode1-3/+1
2007-04-02add a description to viewRay Strode2-4/+9
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-14 04:51:56 +0000