Return to fixedbugs CVS log

Up to [Research Unix] / researchv8dc / cmd / sed

researchv8 Dan Cross

Since V7:

1) garbage could be compiled into pseudo-sense.
Fix: in sed0.c, function compile, case '\0':
remove continue statement, so the '\0' and '\n'
are treated identically.

2) cc warning messages quieted by adding r[12] specifiers
to all references to union reptr.

3) 2/28/80: H command used a null pointer sometimes.
Fix: initialize hspend = holdsp in main().

4) 10/23/80: Null RE could not be first address of range.
Fixed by moving test for null RE into else after test for
no address (near 160 in sed0.c).

5) 2/8/82: got rid of two warning messages from the compiler.
(Changed rep->negfl to rep->r1.negfl in sed0.c;
ipc->r1.lb1 to ipc->r2.lb1 in sed1.c.)

6) 12/6/82: removed (commented out) debugging code in
sed0.c for unrecognized command. (It had been there since
at latest Jan 30, 1981.)

7) 6/22/83: changed the substitute() function in sed1.c
to fix bugs concerning matches of the empty string.
(The previous version would loop.  Note that /bin/ed
as of today has a different bug relating to s/a*b/x/g;
/usr/rob/bin/qed has a third bug.)
s/a*/x/g should now work properly.

8) 4/12/84: added sed: prefix to all error messages.

9) 6/25/84: Removed an artificial limit on the size of a regular expression
by undefinig ESIZE, and making the test of how big a regular expression
can be always be against reend (= &respace[RESIZE].
This limits the total space for storing re's, instead of limiting them
individually.

10) 4/26/85: Increased the size of PTRSPACE (256->512) and RESPACE
(5000->10000).

11) 5/17/85: fixed a bug found by boulder!geoff;
resets sflag (saying whether a substitution was done on the
current line) in routine gline().
Fixes "t" to work as advertized.

12) 7/24/85: changed the 's' case in fcomp to notice if an empty
RE is the first re found.

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.