refactoring some scripts #1

prvst · 2014-09-20T13:02:38Z

No description provided.

Fixups for morse

@lines

* Standard ed treats the address 0 as valid for both "a" and "i" commands [1] * The meaning of address 0 is, add text at start of buffer * When starting ed with no file argument, buffer is empty and current address is 0 * This version of ed accepts "0a" (or just "a") on empty buffer * "0i" command results in error: Can't insert before line 0 * When the buffer is not empty, GNU ed treats 0i the same as 1i * shift(@lines) in edInsert() is safe if the list is empty, so this patch treats address 0 and 1 the same for $INSERT_MODE (aka "i") briandfoy#1 https://www.gnu.org/software/ed/manual/ed_manual.html#Commands Existing output for "a": $ perl ed # a == 0a ME: a ME: hey ME: . ME: 1 ED: hey ME: Q New output for "i": $ perl ed # i == 0i ME: i ME: yo ME: . ME: 1 ED: yo ME: Q

* When viewing text in less(1) and editing with vi(1) I usually enable line numbers * Standard ed supports the n command [1], which prints a tab after a line number prefix * Implement the n command by adding a parameter to edPrint() * Note: "l" is still not implemented but could be a future change to edPrint() briandfoy#1 https://pubs.opengroup.org/onlinepubs/009604599/utilities/ed.html

* When temporarily enabling warnings in bc I observe warnings for incorrect use of "next" * Function yy_err_recover() jumped to a label outside of itself, within yyparse() * yyparse() was strucured so that it first initialises flags then enters main loop * Jumping back to yyparse() was unsafe because the flags would all be cleared again, including $yyssp (yy_err_recover() already adjusts $yyssp) * Clear flags before calling yyparse(), and let yyparse() enter the loop immediately so that it's safe to call it from yy_err_recover() %perl bc # before patch 4++ line 1: syntax error Exiting subroutine via next at bc line 1387, <STDIN> line 1 (briandfoy#1)

* When temporarily enabling warnings in bc I observe warnings for incorrect use of "next" * Function yy_err_recover() jumped to a label outside of itself, within yyparse() * yyparse() was strucured so that it first initialises flags then enters main loop * Jumping back to yyparse() was unsafe because the flags would all be cleared again, including $yyssp (yy_err_recover() already adjusts $yyssp) * Clear flags before calling yyparse(), and let yyparse() enter the loop immediately so that it's safe to call it from yy_err_recover() %perl bc # before patch 4++ line 1: syntax error Exiting subroutine via next at bc line 1387, <STDIN> line 1 (#1)

* patch failed with a strictness error when I specified a file * Patch::apply() calls back into Patch::Context::apply() on L714 with the wrong params * When setting a breakpoint in Patch::apply(), its params are not identical to Patch::Context::apply(), so passing @_ as parameter fails * Reading the code, this would only be a problem for applying context diffs (no callback to self->apply() happened for the other diff types) * I confirmed that this error was not caused by the recent use-base.diff (now I suspect it goes back to 1999) * Removing the code with the incorrect self->apply() call causes patch to correctly report that applying Hunk briandfoy#1 failed, then the program exits with status 1 to indicate that >=1 lines could not be applied

* patch failed with a strictness error when I specified a file * Patch::apply() calls back into Patch::Context::apply() on L714 with the wrong params * When setting a breakpoint in Patch::apply(), its params are not identical to Patch::Context::apply(), so passing @_ as parameter fails * Reading the code, this would only be a problem for applying context diffs (no callback to self->apply() happened for the other diff types) * I confirmed that this error was not caused by the recent use-base.diff (now I suspect it goes back to 1999) * Removing the code with the incorrect self->apply() call causes patch to correctly report that applying Hunk #1 failed, then the program exits with status 1 to indicate that >=1 lines could not be applied

* Remove note about GNU patch requiring a space between -D and argument (this might've been the case for ancient versions but not anymore) * This version of patch outputs a trailing comment for the ifndef blocks; this matches the OpenBSD version %ifconfig > A %sed 's/eth0/ent0/' A > B %perl diff -u A B > AB.diff %/usr/bin/patch --version GNU patch 2.7.6 %/usr/bin/patch -Dostuff -o C A AB.diff patching file C (read from A) %perl patch -Dostuff -o D A AB.diff Hmm... Looks like a unified diff to me... Patching file A using Plan B... Hunk briandfoy#1 succeeded at 1. done %perl diff C D 5c5 < #endif --- > #endif /* ostuff */

* Remove note about GNU patch requiring a space between -D and argument (this might've been the case for ancient versions but not anymore) * This version of patch outputs a trailing comment for the ifndef blocks; this matches the OpenBSD version %ifconfig > A %sed 's/eth0/ent0/' A > B %perl diff -u A B > AB.diff %/usr/bin/patch --version GNU patch 2.7.6 %/usr/bin/patch -Dostuff -o C A AB.diff patching file C (read from A) %perl patch -Dostuff -o D A AB.diff Hmm... Looks like a unified diff to me... Patching file A using Plan B... Hunk #1 succeeded at 1. done %perl diff C D 5c5 < #endif --- > #endif /* ostuff */

* When processing an "a" (append text) command in an ed-style diff, line number ranges are not parsed correctly * If a range (e.g. "1,2") is given as a prefix to "a", the 2nd address ("2") is used as the target address * Regular diff tools don't output "a" commands with 2 addresses * ed-style diffs can be pasted from a previous ed editor session or typed manually, so having extra validation here is helpful * To test this, I bypassed the code for running an external ed program * I modified a diff generated by "diff -e", changing the "322a" command to "1,322a"... the file is patched successfully %cat bad.diffe 1,322a my $Maxlen = 1; # longest string for current directory . 317,318d 113d %perl patch -e ls bad.diffe Hmm... Looks like an ed diff to me... Patching file ls using Plan A... Hunk briandfoy#1 succeeded at 113 Hunk briandfoy#2 succeeded at 317 Hunk briandfoy#3 succeeded at 322 done %cmp ls ls2

* When processing an "a" (append text) command in an ed-style diff, line number ranges are not parsed correctly * If a range (e.g. "1,2") is given as a prefix to "a", the 2nd address ("2") is used as the target address * Regular diff tools don't output "a" commands with 2 addresses * ed-style diffs can be pasted from a previous ed editor session or typed manually, so having extra validation here is helpful * To test this, I bypassed the code for running an external ed program * I modified a diff generated by "diff -e", changing the "322a" command to "1,322a"... the file is patched successfully %cat bad.diffe 1,322a my $Maxlen = 1; # longest string for current directory . 317,318d 113d %perl patch -e ls bad.diffe Hmm... Looks like an ed diff to me... Patching file ls using Plan A... Hunk #1 succeeded at 113 Hunk #2 succeeded at 317 Hunk #3 succeeded at 322 done %cmp ls ls2

* I generated a Context diff with 2 hunks * I manually swapped the 2 numbers at the top of hunk1, from "1,4" to "4,1" (the line number range for the hunk) * GNU patch correctly rejected this invalid diff, but this version crashed with an ARRAY error * Add validation for backwards range, based on code previously added for Normal diffs %cat diff.ctx *** id Wed Mar 5 09:16:33 2025 --- id2 Wed Mar 5 09:14:36 2025 *************** *** 4,1 **** - #!/usr/bin/perl =begin metadata --- 1,3 ---- *************** *** 12,24 **** =cut - # - # An implementation of the 'id' utility in Perl. Written for the Perl Power - # Tools (PPT) project by Theo Van Dinter (felicity@kluge.net). - # - # $Id: id,v 1.2 2004/08/05 14:17:43 cwest Exp $ - # - use strict; use File::Basename qw(basename); --- 11,16 ---- %perl -Mwarnings -Mdiagnostics patch -c id diff.ctx Hmm... Looks like a context diff to me... Patching file id using Plan A... Short hunk ignored...no second line range. 1 out of 1 hunks ignored--saving rejects to id.rej Can't use an undefined value as an ARRAY reference at patch line 382, <$in_fh> line 249 (briandfoy#1) (F) A value used as either a hard reference or a symbolic reference must be a defined value. This helps to delurk some insidious errors. Uncaught exception from user code: Can't use an undefined value as an ARRAY reference at patch line 382, <$in_fh> line 249. Patch::bless(Patch::Context=HASH(0x55908a4f50)) called at patch line 161

* I generated a Context diff with 2 hunks * I manually swapped the 2 numbers at the top of hunk1, from "1,4" to "4,1" (the line number range for the hunk) * GNU patch correctly rejected this invalid diff, but this version crashed with an ARRAY error * Add validation for backwards range, based on code previously added for Normal diffs %cat diff.ctx *** id Wed Mar 5 09:16:33 2025 --- id2 Wed Mar 5 09:14:36 2025 *************** *** 4,1 **** - #!/usr/bin/perl =begin metadata --- 1,3 ---- *************** *** 12,24 **** =cut - # - # An implementation of the 'id' utility in Perl. Written for the Perl Power - # Tools (PPT) project by Theo Van Dinter (felicity@kluge.net). - # - # $Id: id,v 1.2 2004/08/05 14:17:43 cwest Exp $ - # - use strict; use File::Basename qw(basename); --- 11,16 ---- %perl -Mwarnings -Mdiagnostics patch -c id diff.ctx Hmm... Looks like a context diff to me... Patching file id using Plan A... Short hunk ignored...no second line range. 1 out of 1 hunks ignored--saving rejects to id.rej Can't use an undefined value as an ARRAY reference at patch line 382, <$in_fh> line 249 (#1) (F) A value used as either a hard reference or a symbolic reference must be a defined value. This helps to delurk some insidious errors. Uncaught exception from user code: Can't use an undefined value as an ARRAY reference at patch line 382, <$in_fh> line 249. Patch::bless(Patch::Context=HASH(0x55908a4f50)) called at patch line 161

Felipe da Veiga Leprevost added 2 commits September 20, 2014 10:01

substituting require directives for use in mose

81ab57d

fixing ping indentation and removing defined command

ce489b8

prvst changed the title ~~substituting require directives for use in mose~~ substituting require directives for use in morse Sep 20, 2014

prvst changed the title ~~substituting require directives for use in morse~~ refactoring some scripts Sep 20, 2014

briandfoy added a commit that referenced this pull request Sep 20, 2014

Merge pull request #1 from Leprevost/master

608e582

Fixups for morse

briandfoy merged commit 608e582 into briandfoy:master Sep 20, 2014

briandfoy added Program: morse The morse program Type: modernization updating programs to current practices Program: ping The ping program Status: released there is a new release with this fix labels Jul 11, 2023

briandfoy self-assigned this Jul 11, 2023

mknos mentioned this pull request Aug 7, 2024

patch: runtime error when patching the wrong file #703

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

refactoring some scripts #1

refactoring some scripts #1

Uh oh!

prvst commented Sep 20, 2014

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

refactoring some scripts #1

refactoring some scripts #1

Uh oh!

Conversation

prvst commented Sep 20, 2014

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants