summaryrefslogtreecommitdiff
path: root/xc
AgeCommit message (Collapse)AuthorFilesLines
1995-06-26XBUG 7483kaleb1-4/+4
1995-06-26XBUG 7024kaleb1-1/+3
1995-06-26add DBEdpw1-2/+12
1995-06-23Reimplement XtGetSelectionParameters. Find the appropriate propertyconverse1-45/+58
and the requestor's window through the Request record, either passed (if the owner is incremental) or discovered through the owner's Select context. Reuse code from XtGetSelectionRequest. Assume the owner is not dispatching from the conversion procedure in referencing the Request record through the owner context record. #7028 No need for the requestor to wrap calls to set a property on its own window with an error handler.
1995-06-22Change how the ApplicationShell treats argv. It is copied, and itsconverse1-10/+46
contents are copied, and it is not assumed to be ended by a NULL element. An ending NULL element is present in the copy. Allow -1 as argc to indicate ignorning argv. Assume caller treats argc and argv as a resource pair that do not contain conflicting values. #7157
1995-06-22no bug-report, for R7. JapanTaiUKUS4Modmap was the same as the Generic5Modmapkaleb1-22/+5
so use it instead of carrying a duplicate redundant modmap in the source.
1995-06-22XBUG 7479. Eliminate warning for bad castkaleb1-2/+2
1995-06-22XBUG 7302 supplement. Even with the prior change, the chances of thekaleb1-20/+20
cache being rendered invalid were high due to the way the modmix table was seeded. When the modifier combinations have hight order bits set the hash results when subsequent combinations without the high order bits is the same and the cache entry will be overwritten. This results in the key-translator being called more times than is necessary. By seeding the modmix table differently it's possible to arrange for the hash to produce a different index for modifier combinations with the high bit set and we see a significant reduction in the number of times that the key-translator is called in most cases.
1995-06-22XBUG 7302. The loop in _XtMatchUsingDontCareMods was a lazy way tokaleb1-17/+62
try all the combinations of useful modifiers and it had the undesirable side effect of rendering the key-translator cache invalid. This happens because the hash for a combination with a high bit set results in the same value as when the high bit is not set. Given the size of the cache and other factors when you test extra bit combinations that are not valid combinations of the useful_mods the result is that the cache entry for a given key is routinely overwritten by useless modifier combinations. By only testing real combinations of the useful modifiers the chance of overwriting the cache and rendering it invalid is reduced somewhat.
1995-06-19del dup XConsortium linekaleb2-4/+2
1995-06-19add XConsortium linekaleb1-0/+1
1995-06-19add XConsortium linekaleb2-0/+2
1995-06-19add XConsortium linekaleb2-0/+2
1995-06-19Add XConsortium linekaleb1-0/+1
1995-06-19Add XConsortium linekaleb1-0/+1
1995-06-19eliminate dup XConsortium linekaleb1-2/+1
1995-06-19delete extraneous XConsortium linekaleb4-8/+4
1995-06-19define HasLdRunPath. Close XBUG #7475.gildea1-1/+4
1995-06-16no bug-report, Solaris 2.4 on Intelkaleb1-3/+3
1995-06-16no bug-report, misc improvements for SGI, e.g. SVR4-style shared libskaleb2-4/+6
on IRIX 5 and IRIX 6.
1995-06-16add information about make variables that can be set in an Imakefilegildea1-1/+227
and add lots of general advice about writing an Imakefile.
1995-06-16XBUG 7187. BlockHook might add or remove file descriptors. The blockkaleb3-18/+48
hook might also generate protocol. Need to check correctly for both cases. As a byproduct, if workprocs or signal handlers do either they will be properly noticed now too.
1995-06-16no bug-report, error in prior edit for installing libraries.kaleb1-2/+2
1995-06-16ToolkitMakeStringsRule -> ToolkitMakeStringsgildea1-2/+5
with an added argument listing template files to depend on
1995-06-16ToolkitMakeStringsRule -> ToolkitMakeStringsgildea1-21/+24
with an added "depends" argument.
1995-06-16XBUG 7474, make certain to correctly generate the intervening ',' whenkaleb1-7/+17
-earlyR6bc is specified. Without it the resulting source file won't compile.
1995-06-15It is not necessary to cache the display along with a cache of widgets,converse1-11/+4
as widget addresses are unique across multiple displays. This brings the pseudoTrace cache into the same structure as the pathTrace cache, which was changed in the previous revision. Ship with fix for #7024
1995-06-15In the widget destroy callback which cleans up the per widget input record,converse1-1/+2
look at the cache used to hold the focus source widget and its ancestors. Clear it if the cache contains the widget about to be destroyed. This change is linked to changes in Keyboard.c #7024
1995-06-15The pathTrace array is a cache of a widget and its ancestors, up to theconverse1-9/+14
shell ancestor, used by InActiveSubtree for Xt focus management. It must be cleared when the widget is destroyed if that widget is in the cache, so that a widget created in the future with the same address will not be confused with the stale pointer in the cache. There is no need to include the display of the widget in the cached info. #7024
1995-06-15In R6, due to the hook registration widget, it is no longer true thatconverse1-3/+2
a widget which is not a shell must have a parent. #7473
1995-06-15fix XBUG 7468: missing parens in macro definitionsdpw1-3/+3
1995-06-15document format of MIT-KERBEROS-5 authority info passed to server.gildea1-21/+34
1995-06-15While walking over an uncomplied resource list in _XtTypedArgToArg,converse1-9/+6
use strcmp instead of doing quark comparisions, as optimization. #7466
1995-06-15fix XBUG 7007: lines sometimes drawn as trianglesdpw1-2/+2
1995-06-15close XBUG 5782: miPolyArc can be *very* inefficient at drawing small ↵dpw1-29/+45
sections of arcs for circles of very large radius.
1995-06-15add DBEdpw1-3/+7
1995-06-15fix XBUG 6940: missing .EXIT before .PROCENDdpw1-1/+2
1995-06-15delete Apple, ATR, DRS, GATech, JCC, Locus, M3i, ParcPlace, Phase Xrws1-10/+1
1995-06-14XBUG 7467, do the right thing for functions that need an ANSI-stylekaleb1-1/+5
declaration when there's a prototype in scope (there is)
1995-06-14XBUG 7467, do the right thing for functions that need an ANSI-style declarationkaleb5-5/+25
when there's a prototype in scope (there is)
1995-06-14XBUG 7189, eliminate the remainders of the undocumented INTERNAL selectionkaleb4-86/+11
mechanism that was added along with the i18n support
1995-06-10add DBEdpw1-1/+7
1995-06-09change #if __STDC__ to #ifdef __STDC__gildea2-4/+4
so can pass -Xa instead of -Xc to SVR4 compilers. See advisory message of 21 April 1995.
1995-06-09change #if __STDC__ to #ifdef __STDC__gildea3-6/+6
so can pass -Xa instead of -Xc to SVR4 compilers. See advisory message of 21 April 1995.
1995-06-09change #if __STDC__ to #ifdef __STDC__gildea1-3/+3
so can pass -Xa instead of -Xc to SVR4 compilers. See advisory message of 21 April 1995.
1995-06-09no bug report, NEWS-OS 6.1.0kaleb1-4/+4
1995-06-09change #if __STDC__ to #ifdef __STDC__gildea12-37/+21
so can pass -Xa instead of -Xc to SVR4 compilers. See advisory message of 21 April 1995.
1995-06-08change #if __STDC__ to #ifdef __STDC__gildea61-163/+168
so can pass -Xa instead of -Xc to SVR4 compilers. See advisory message of 21 April 1995.
1995-06-08Do tty setting correctly on SVR4-like machines. XBUGs #7051, #7085.gildea1-76/+216
1995-06-08A new function, _XtFreeArgList, is called from XtVaSetValues. Thisconverse1-13/+73
function frees the argument list, which doubles as a data structure to represent memory allocated to hold a value returned in a typed argument conversion. This overloading of the use of the argument list is inelegant but it preserves the interface of _XtVaToArgList. #7167