| Geth | rakudo: ugexe++ created pull request #6630: RakuAST: fix several compile-time variable constructs |
00:41 | |
| roast: 48148911ee | (Nick Logan)++ (committed using GitHub Web editor) | S29-os/system.t Run a silent child in the RT #115390 loops (#899) The 200 iterations ran raku -v with inherited standard output, so every child wrote a three line version banner into the harness stream. Under CI load a banner chunk can land inside a test line, which corrupts the numbering and fails the file with tests out of sequence. An empty -e program spawns the same way and prints nothing. The loops only pin that run() in a loop does not crash, in sink context and assigned, so the command output was never part of the test. |
00:54 | ||
| roast: 3382e30698 | (Nick Logan)++ (committed using GitHub Web editor) | S32-io/IO-Socket-Async.t Make the EADDRINUSE quit test deterministic (#898) Previously this test picked a random port between 5000 and 10000 and tapped the same listen supply twice, expecting the first tap to win the bind and the second to quit. When something on the machine already held that port both listens failed. The first tap had no quit handler, and a failed listen also panics MoarVM on the next listen tap cancellation, which killed the test file after test 26. This binds a holder listener on port 0 first so the port is guaranteed to be in use, taps a single listener on that port with a quit handler, and checks the promise status rather than the always-truthy promise object. |
00:55 | ||
| linkable6 | ROAST#898 [closed]: github.com/Raku/roast/pull/898 Make the EADDRINUSE quit test deterministic | ||
| roast: 177d7af123 | (Nick Logan)++ (committed using GitHub Web editor) | S17-channel/basic.t Wait for the Channel Supply to drain before checking what it saw (#897) Previously the test closed the channel and immediately checked @seen, but Channel.Supply delivers values to its tap on a thread pool thread. Nothing synchronized the check with that delivery, so under load the main thread could read @seen before the tap finished pushing the last values. This taps with a done callback and awaits its promise after closing the channel, which Channel.Supply keeps only once the channel is drained. |
|||
| linkable6 | ROAST#897 [closed]: github.com/Raku/roast/pull/897 Wait for the Channel Supply to drain before checking what it saw | ||
| roast/ugexe/xor-warning-line: 2a58ec1d55 | (Nick Logan)++ | 4 files Reach the run time type check with a value the compiler cannot judge Rakudo is learning to report a store of a value a variable's type can never accept while it compiles, instead of leaving it to the run time check. A test written as `dies-ok { my Int $x = "str" }` names a literal the compiler can judge, so it stops reaching the check it was written for. ... (6 more lines) |
00:56 | ||
| roast/ugexe/xor-warning-line: 455d32022f | (Nick Logan)++ | S03-operators/short-circuit.t Judge the xor precedence warning by its own line Previously the three assertions searched the whole of stderr for the operand the warning is about and for the absence of the others. A compiler that quotes the source under a warning prints every operand there, so the search found the excluded ones in the excerpt rather than in the warning. This looks only at the lines carrying the warning, which is what the test was written to judge. |
|||
| roast/ugexe/xor-warning-line: 82fcecf41c | (Nick Logan)++ | S03-operators/short-circuit.t Judge the xor precedence warning by its own line Previously the three assertions searched the whole of stderr for the operand the warning is about and for the absence of the others. A compiler that quotes the source under a warning prints every operand there, so the search found the excluded ones in the excerpt rather than in the warning. This looks only at the lines carrying the warning, which is what the test was written to judge. |
00:58 | ||
| roast/ugexe/xor-warning-line: 17f760dd7c | (Nick Logan)++ | S03-operators/short-circuit.t Judge the xor precedence warning by its own line Previously the three assertions searched the whole of stderr for the operand the warning is about and for the absence of the others. A compiler that quotes the source under a warning prints every operand there, so the search found the excluded ones in the excerpt rather than in the warning. This looks only at the lines carrying the warning, which is what the test was written to judge. |
00:59 | ||
| roast: ugexe++ created pull request #902: Judge the xor precedence warning by its own line |
|||
| roast/ugexe/literal-store-compile-time: a82b8bc0d4 | (Nick Logan)++ | 2 files Hold the rejected string in one variable Previously the native declarator tests held their string in a variable of their own beside the one the typed declarator tests use. This has every test store the same string. |
01:01 | ||
| roast: 74193e17bd | (Nick Logan)++ | 4 files Reach the run time type check with a value the compiler cannot judge Rakudo is learning to report a store of a value a variable's type can never accept while it compiles, instead of leaving it to the run time check. A test written as `dies-ok { my Int $x = "str" }` names a literal the compiler can judge, so it stops reaching the check it was written for. ... (6 more lines) |
01:02 | ||
| roast: a82b8bc0d4 | (Nick Logan)++ | 2 files Hold the rejected string in one variable Previously the native declarator tests held their string in a variable of their own beside the one the typed declarator tests use. This has every test store the same string. |
|||
| roast: 18a00054c7 | (Nick Logan)++ (committed using GitHub Web editor) | 4 files Merge pull request #900 from Raku/ugexe/literal-store-compile-time Reach the run time type check with a value the compiler cannot judge |
|||
| roast: c65cc7a8de | (Nick Logan)++ (committed using GitHub Web editor) | S03-operators/short-circuit.t Judge the xor precedence warning by its own line (#902) Previously the three assertions searched the whole of stderr for the operand the warning is about and for the absence of the others. A compiler that quotes the source under a warning prints every operand there, so the search found the excluded ones in the excerpt rather than in the warning. This looks only at the lines carrying the warning, which is what the test was written to judge. |
|||
| linkable6 | ROAST#902 [closed]: github.com/Raku/roast/pull/902 Judge the xor precedence warning by its own line | ||
| roast/ugexe/6.c-errata-sync: 3312f0629f | (Nick Logan)++ | 3 files Reach the run time type check with a value the compiler cannot judge Rakudo is learning to report a store of a value a variable's type can never accept while it compiles, instead of leaving it to the run time check. A test written as `dies-ok { my Int $x = "str" }` names a literal the compiler can judge, so it stops reaching the check it was written for. ... (6 more lines) |
01:12 | ||
| roast/ugexe/6.c-errata-sync: ae21fc7cad | (Nick Logan)++ | S04-declarations/my.t Hold the rejected string in one variable Previously the native declarator tests held their string in a variable of their own beside the one the typed declarator tests use. This has every test store the same string. |
|||
| roast/ugexe/6.c-errata-sync: 61c0a329f9 | (Nick Logan)++ | S29-os/system.t Run a silent child in the RT #115390 loops (#899) The 200 iterations ran raku -v with inherited standard output, so every child wrote a three line version banner into the harness stream. Under CI load a banner chunk can land inside a test line, which corrupts the numbering and fails the file with tests out of sequence. An empty -e program spawns the same way and prints nothing. The loops only pin that run() in a loop does not crash, in sink context and assigned, so the command output was never part of the test. |
|||
| roast/ugexe/6.d-errata-sync: 5 commits pushed by (Nick Logan)++ | |||
| roast: ugexe++ created pull request #903: Sync recent roast adjustments to 6.c-errata |
|||
| roast: ugexe++ created pull request #904: Sync recent roast adjustments to 6.c-errata |
|||
| rakudo: ugexe++ created pull request #6631: RakuAST: Parse -M arguments as use statements |
01:51 | ||
|
02:03
japhb left
|
|||
| Geth | rakudo/main: d5a78515a6 | (Nick Logan)++ (committed using GitHub Web editor) | 13 files RakuAST: Show the source around a compile time problem (#6629) Previously only the grammar's own errors carried the excerpt of source that renders as the `------>` line with the problem marked in it, since the cursor position was at hand when they were built. Every problem the frontend found after parsing rendered with a file and line and nothing of the source: check time sorries and worries, a BEGIN or CHECK block ... (9 more lines) |
02:18 | |
|
02:20
japhb joined
|
|||
| ugexe | so all these years we have been using :auth<...> | 02:56 | |
| but that way doesnt actually work with spaces | 02:57 | ||
| m: use Test:auth<Yet Another Society>; | 02:59 | ||
| camelia | ===SORRY!=== Error while compiling <tmp> Could not find Test:auth<Yet Another Society> in: /home/camelia/.raku /home/camelia/rakudo-m-inst-2/share/perl6/site /home/camelia/rakudo-m-inst-2/share/perl6/vendor /home/camelia/raku… |
||
| ugexe | m: use Test:auth("Yet Another Society"); | ||
| camelia | ( no output ) | ||
|
04:07
vrurg_ left
|
|||