{ "tokens": { "input": 127888, "cache_read": 3029360, "cache_write": 757340, "output": 16460 }, "reviews": [ { "key": [ "https://ftp.debian.org/debian/pool/main/a/apache-arrow/apache-arrow_25.0.0.orig.tar.gz", "your apache-arrow-25.0.0/cpp/src/arrow/util/bitmap_reader_benchmark.cc:41:27: style: int result is assigned to long variable. If the variable is long to avoid loss of information, then you have loss of information. [truncLongCastAssignment]" ], "verdict": { "verdict": "regression", "confidence": "medium", "explanation": "The new warning is a false positive. `rand() % 25` is in 0..24, so `1024 * (rand() % 25 + 1000)` is at most 1024*1024 = 1,048,576. That fits easily in `int`, so assigning the int product to `int64_t` cannot lose information. Main apparently used the bounded value range to suppress `truncLongCastAssignment`. The PR's valueflow changes seem to have lost that range information, producing a new false positive." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/b/bamtools/bamtools_2.5.3+dfsg.orig.tar.xz", "main bamtools-2.5.3+dfsg/src/api/BamAlignment.cpp:154:28: style: Comparing expression of type 'const char' against value 255. Condition is always false. [compareValueOutOfTypeRangeError]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "The code compares `qualData[0] == (char)0xFF`. Both operands are plain `char`, so the right-hand side is the same `char` value that `qualData[0]` can hold: -1 if char is signed, 255 if it is unsigned. The condition is true whenever the byte is 0xFF, so 'always false' was a false positive. Main did not apply the char cast before comparing against the type range. The PR handles plain-char signedness and the cast correctly, so removing the warning is an improvement." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/c/caja/caja_1.26.4.orig.tar.xz", "main caja-1.26.4/libcaja-private/caja-progress-info.c:714:17: style: Condition 'info->waiting' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Both main and the PR report the same knownConditionTrueFalse at line 714. Only the attached notes (the error path) differ, so this is not a removal. The code is a double-checked locking pattern: 'info->waiting' is checked again after G_LOCK because another thread may have changed it. The warning is therefore a likely false positive in both versions. The PR neither removes nor fixes it; it only changes how the warning is presented, which does not matter to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/c/contourpy/contourpy_1.3.3.orig.tar.gz", "main contourpy-1.3.3/src/mpl2005_original.cpp:569:32: style: Condition 'level0' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "The same knownConditionTrueFalse warning at line 569 appears in both runs. Inside the `if (level0 && ...)` block, `level0` is indeed always true, so the warning is a true positive either way. The main/your pair only reflects a difference in the error path notes, such as \"Assuming condition is false\" for line 564, which was present in the your output. The location and message are unchanged, so this does not matter to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/c/coreutils/coreutils_9.12.orig.tar.xz", "your coreutils-9.12/lib/fnmatch_loop.c:925:38: style: Condition 'n!=string_end' is always true [knownConditionTrueFalse]\n coreutils-9.12/lib/fnmatch_loop.c:922:9: note: Assuming that condition 'n==string_end' is not redundant\n coreutils-9.12/lib/fnmatch_loop.c:925:38: note: Condition 'n!=string_end' is always true" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "The same warning is reported by both main and the PR. Only the error-path notes differ: the PR now shows the note 'Assuming that condition n==string_end is not redundant', so the line appears as one main and one your entry. The warning itself is a true positive. Line 922 returns when n == string_end, so on line 925 'n != string_end' is always true (it is redundant). The added note points at the earlier check that makes the condition redundant, which may help users slightly, but the change does not really matter either way." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/d/dcap/dcap_2.47.14.orig.tar.gz", "your dcap-2.47.14/src/dcap_rename.c:65:10: style: Condition 'url==NULL' is always false [knownConditionTrueFalse]\n dcap-2.47.14/src/dcap_rename.c:49:10: note: Assuming that condition 'url==NULL' is not redundant\n dcap-2.47.14/src/dcap_rename.c:57:11: note: Assuming condition is false\n dcap-2.47.14/src/dcap_rename.c:65:10: note: Condition 'url==NULL' is always false" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Line 49 returns early when `url == NULL`, and `url` is not changed afterwards, so `url == NULL` at line 65 really is always false. Both versions report this true positive with the same message. Only the attached note path differs: the new one adds an 'Assuming condition is false' note for the `node == NULL` check at line 57. That note is harmless and does not change how useful or accurate the warning is." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/d/ddcci-driver-linux/ddcci-driver-linux_0.4.5.orig.tar.gz", "your ddcci-driver-linux-v0.4.5/ddcci-backlight/ddcci-backlight.c:163:7: style: Condition '!ptr' is always false [knownConditionTrueFalse]\n ddcci-driver-linux-v0.4.5/ddcci-backlight/ddcci-backlight.c:162:14: note: Assuming that condition 'ptr=strpbrk(ptr,\" ()\")' is not redundant\n ddcci-driver-linux-v0.4.5/ddcci-backlight/ddcci-backlight.c:163:7: note: Condition '!ptr' is always false" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Inside `while ((ptr = strpbrk(...)))` the check `!ptr` on line 163 really is always false, so the warning is a true positive both before and after the PR. The only difference is that the 'your' version adds an extra note, 'Assuming that condition ... is not redundant'. The warning text, line and severity are unchanged, so this makes no practical difference to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/d/deepin-qt5dxcb-plugin/deepin-qt5dxcb-plugin_5.7.12.orig.tar.gz", "main qt5platform-plugins-5.7.12/src/ddesktopinputselectioncontrol.cpp:434:43: style: Condition 'm_anchorSelectionHandle' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "Line 434 tests `m_anchorSelectionHandle && m_anchorSelectionHandle`. The second operand repeats the first (the author probably meant m_cursorSelectionHandle), so it is always true and the warning is a true positive. Both versions report it. The PR only adds an extra note line, 'Assuming that condition ... is not redundant', pointing at the first occurrence. Only the error path changed, and it now shows where the always-true conclusion comes from. That makes the message more informative." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/e/e2guardian/e2guardian_5.5.9.orig.tar.gz", "your e2guardian-5.5.9r/src/IPList.cpp:179:46: warning: Expression 'host_part' can have a negative value. That is converted to an unsigned value and used in an unsigned calculation. [signConversionCond]\n e2guardian-5.5.9r/src/IPList.cpp:178:27: note: Assuming that condition 'host_part>-1' is not redundant\n e2guardian-5.5.9r/src/IPList.cpp:179:46: note: Negative value is converted to an unsigned value" ], "verdict": { "verdict": "regression", "confidence": "high", "explanation": "This new signConversionCond warning is a false positive. Line 179 only runs inside `if (host_part > -1)`, so `host_part` is at least 0 there and cannot be negative. Also, in `0xFFFFFFFF << host_part` the shift count is never converted to unsigned: shift operands are promoted separately, with no usual arithmetic conversion, so `host_part` stays `int`. The warning misreads the guard condition as allowing negative values and wrongly treats the shift count as part of an unsigned calculation." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/e/engrampa/engrampa_1.28.2.orig.tar.xz", "your engrampa-1.28.2/src/fr-command-alz.c:105:6: style: Condition '!alz_comm->list_started' is always false [knownConditionTrueFalse]\n engrampa-1.28.2/src/fr-command-alz.c:93:6: note: Assuming that condition '!alz_comm->list_started' is not redundant\n engrampa-1.28.2/src/fr-command-alz.c:99:34: note: Assuming condition is false\n engrampa-1.28.2/src/fr-command-alz.c:105:6: note: Condition '!alz_comm->list_started' is always false" ], "verdict": { "verdict": "neutral", "confidence": "low", "explanation": "The warning is a true positive in both versions. If `list_started` is false at line 93, the function returns. If the `strncmp` at line 99 matches, it also returns. So at line 105 `list_started` is still true and `!alz_comm->list_started` is always false. Main and your report the same location and message; only the error-path notes differ, and main's notes are not shown. Your path adds 'Assuming condition is false' at 99:34, which is accurate but does not change the diagnosis. The same pattern appears in the other entries for this package, so the change is essentially neutral." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/f/frugen/frugen_3.1.orig.tar.xz", "your frugen-3.1/lib/fru_delete_custom.c:44:10: style: Condition 'rec' is always true [knownConditionTrueFalse]\n frugen-3.1/lib/fru_delete_custom.c:38:6: note: Assuming that condition '!rec' is not redundant\n frugen-3.1/lib/fru_delete_custom.c:44:10: note: Condition 'rec' is always true" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "In `delete_reclist_entry`, line 38 returns early when `rec` is null, so `rec` really is always true at line 44. Both versions report this true positive. The PR's version adds an error path: it points to the `!rec` check at line 38 and then to line 44. That makes the warning easier to understand, so it is a slightly better message." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/g/gkl/gkl_0.9.1+dfsg.orig.tar.xz", "your GKL-0.9.1/src/main/native/compression/isa-l-master/igzip/igzip_rand_test.c:2624:12: style: Condition 'z_buf!=NULL' is always true [knownConditionTrueFalse]\n GKL-0.9.1/src/main/native/compression/isa-l-master/igzip/igzip_rand_test.c:2604:12: note: Assuming that condition 'z_buf==NULL' is not redundant\n GKL-0.9.1/src/main/native/compression/isa-l-master/igzip/igzip_rand_test.c:2612:14: note: Assuming condition is false\n GKL-0.9.1/src/main/native/compression/isa-l-master/igzip/igzip_rand_test.c:2624:12: note: Condition 'z_buf!=NULL' is always true" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Both main and the PR report the same warning at line 2624. It is a true positive: `z_buf == NULL` returns early at line 2604, so `z_buf != NULL` at line 2624 is always true. Only the error-path notes changed. The PR's path now adds an 'Assuming condition is false' note for the unrelated `tmp_buf == NULL` check at line 2612. That note is harmless but adds no useful information. The main line's notes are not shown, so the change cannot be judged clearly better or worse; for users the diagnostic is effectively the same." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/g/glusterfs/glusterfs_11.2.orig.tar.gz", "your glusterfs-11.2/xlators/features/quota/src/quota.c:81:9: style: Condition 'inode' is always true [knownConditionTrueFalse]\n glusterfs-11.2/xlators/features/quota/src/quota.c:78:24: note: Assuming that condition 'inode==NULL' is not redundant\n glusterfs-11.2/xlators/features/quota/src/quota.c:81:9: note: Condition 'inode' is always true" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "Line 78 returns early when `inode == NULL`, so `if (inode)` on line 81 really is always true. This is a true positive. It appears as both a main and a your line with identical text, and the diff lists notes only once for the pair. So the warning is still reported; only the ordering or notes changed. No effect for users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/g/golang-github-wellington-go-libsass/golang-github-wellington-go-libsass_0.9.2+git20181130.4ef5b9d.orig.tar.xz", "main go-libsass/libsass-build/node.cpp:212:11: style: Condition 'pToConvert' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "The same warning, 'Condition pToConvert is always true' at node.cpp:212, is reported by both main and your. Only the note lines attached to the 'your' entry differ, which is why it appears as a main/your pair. The warning is correct: line 212 is inside `while (pToConvert)`, and pToConvert is not modified before line 212, so the check is redundant. The finding itself did not change, so the result does not matter to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/h/hashrat/hashrat_1.25+ds.orig.tar.xz", "main ColumPaget-Hashrat-10c0432/libUseful-bundled/RawData.c:195:12: style: Condition 'pos==-1' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "`pos` is a size_t. In `pos==-1` the -1 is converted to SIZE_MAX, so the comparison is true whenever the caller passes -1, which is a common sentinel idiom. Main wrongly said the condition is always false, presumably because it ignored the unsigned conversion of -1 and treated the size_t as never negative. The PR adds proper common-type and implicit-conversion handling, so this false positive is removed." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/lib3/lib3mf/lib3mf_1.8.1+ds.orig.tar.xz", "your lib3mf-1.8.1+ds/Source/Common/Platform/NMR_XmlReader_Native.cpp:900:14: style: Condition 'pAmp!=nullptr' is always true [knownConditionTrueFalse]\n lib3mf-1.8.1+ds/Source/Common/Platform/NMR_XmlReader_Native.cpp:896:18: note: Assuming that condition 'pAmp==nullptr' is not redundant\n lib3mf-1.8.1+ds/Source/Common/Platform/NMR_XmlReader_Native.cpp:899:26: note: Assuming condition is true\n lib3mf-1.8.1+ds/Source/Common/Platform/NMR_XmlReader_Native.cpp:900:14: note: Condition 'pAmp!=nullptr' is always true" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "The warning at line 900 is a true positive in both versions. Line 900 is reached only through the `else` of `pAmp == nullptr` on line 896, so `pAmp != nullptr` is always true there. Main and the PR both report it. Only the error path notes differ: the PR adds an 'Assuming condition is true' step for line 899. The location, message and correctness are unchanged, so this does not matter to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/lib3/lib3mf/lib3mf_1.8.1+ds.orig.tar.xz", "your lib3mf-1.8.1+ds/Source/Model/COM/NMR_COMInterface_Slice.cpp:66:27: style: Condition 'pSliceStackResource==nullptr' is always false [knownConditionTrueFalse]\n lib3mf-1.8.1+ds/Source/Model/COM/NMR_COMInterface_Slice.cpp:62:27: note: Assuming that condition 'pSliceStackResource==nullptr' is not redundant\n lib3mf-1.8.1+ds/Source/Model/COM/NMR_COMInterface_Slice.cpp:66:27: note: Condition 'pSliceStackResource==nullptr' is always false" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "Line 66 rechecks 'pSliceStackResource == nullptr' right after line 62 already threw on null, and the pointer was dereferenced in between. So the condition really is always false; it is a copy-paste bug where 'pSliceStack' was meant. Main reported the same warning without a supporting error path. The PR keeps the true positive and adds notes pointing to the earlier null check at line 62, which makes the message more informative." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/libc/libctapimkt/libctapimkt_1.0.5~rc1.orig.tar.xz", "main libctapimkt-1.0.5~rc1/ctapi.c:170:13: style: Condition 'flags==-1' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Line 168 already does `if(flags==-1) err(1,NULL);`, and `err` does not return. So at line 170 `flags` cannot be -1, and the repeated check is truly always false. This is a true positive in both versions; the pull request only changed the attached notes. The new path adds an extra note at 169:45, 'Assuming condition is false' about `fcntl(...)==-1`. That note is harmless but irrelevant, since the conclusion does not depend on it. The warning text, location and correctness are unchanged, so this matters little to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/libe/libemf2svg/libemf2svg_1.8.1+ds.orig.tar.xz", "your libemf2svg-1.8.1/src/lib/emf2svg_utils.c:1234:14: style: Condition 'rcmap.uni' is always true [knownConditionTrueFalse]\n libemf2svg-1.8.1/src/lib/emf2svg_utils.c:1229:19: note: Assuming that condition 'rcmap.uni==NULL' is not redundant\n libemf2svg-1.8.1/src/lib/emf2svg_utils.c:1234:14: note: Condition 'rcmap.uni' is always true" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "The same warning is reported by both versions. Only the attached notes differ: the PR adds the 'Assuming that condition rcmap.uni==NULL is not redundant' note pointing to line 1229. The warning itself is a true positive. After the early return at line 1229 when rcmap.uni is NULL, rcmap.uni is always non-NULL at line 1234. The added note explains where the value comes from, which helps a little but does not change the diagnostic. Effectively neutral, at most a slight improvement in message context." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/libf/libfcgi/libfcgi_2.4.5.orig.tar.gz", "your fcgi2-2.4.5/cgi-fcgi/cgi-fcgi.c:894:8: style: Condition 'exitStatusSet' is always true [knownConditionTrueFalse]\n fcgi2-2.4.5/cgi-fcgi/cgi-fcgi.c:887:11: note: Assuming that condition '!exitStatusSet' is not redundant\n fcgi2-2.4.5/cgi-fcgi/cgi-fcgi.c:894:8: note: Condition 'exitStatusSet' is always true" ], "verdict": { "verdict": "neutral", "confidence": "low", "explanation": "The line numbers in the excerpt are off by about 12. The real code is `while(!exitStatusSet) { OS_DoIo(NULL); } if(exitStatusSet) ...`. Once the loop exits, `exitStatusSet` must be non-zero, so the warning that `exitStatusSet` is always true is a true positive. Both main and the PR report it at the same location with the same message. Only the attached notes/error path differ. The PR's notes ('Assuming that condition !exitStatusSet is not redundant' -> 'always true') correctly explain it, but main's notes are not shown, so there is no evidence the message became clearly more or less accurate." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/libn/libnjb/libnjb_2.2.7~dfsg0.orig.tar.bz2", "main libnjb-2.2.7~dfsg0/sample/fwupgrade.c:286:13: style: Condition 'bread==-1' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "`bread` holds the result of `read_in_fw()`, which in libnjb returns -1 when opening or reading the file fails. Even if `bread` is an unsigned type such as `size_t`, the -1 is converted to the maximum value on assignment. In `bread == -1` the -1 literal gets the same conversion, so the comparison can be true. Calling it 'always false' was a false positive, probably from the old treatment of unsigned-versus-negative comparisons. The PR now models the usual arithmetic conversions correctly, so dropping the warning is an improvement. The declaration of `bread` is not shown, which limits confidence." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/libs/libsoldout/libsoldout_1.4.orig.tar.bz2", "main libsoldout-1.4/mkd2man.c:143:6: style: Condition 'text' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "After `if (!text || !text->size) return 0;`, `text` is known to be non-null, so `if (text)` on line 143 really is always true. The warning is a true positive in both versions. The main and your lines report the same warning at the same location with the same text. The only difference is the note lines attached to the 'your' entry, which most likely is just how the diff shows the error path. It is not a real change for users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/libs/libspctag/libspctag_0.2.orig.tar.gz", "your libspctag-0.2/src/spctag.c:380:19: style: Condition 'fadelength>0xFFFFFFFF' is always false [knownConditionTrueFalse]\n libspctag-0.2/src/spctag.c:378:25: note: Assignment 'fadelength=(unsigned int)atoi(new_fadelength)', assigned value is less than 4294967296\n libspctag-0.2/src/spctag.c:380:19: note: Condition 'fadelength>0xFFFFFFFF' is always false" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "`fadelength` is a 32-bit `unsigned`, so `fadelength > 0xFFFFFFFF` can never be true. Main reported this as compareValueOutOfTypeRangeError. With the PR, the cast gives the value a range below 4294967296, the comparison gets a known false value, and the new code skips the type-range warning to avoid a duplicate. The same true positive is now reported as knownConditionTrueFalse with an accurate value-flow note. The finding is equally correct and useful either way; only the checker id and wording changed." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/m/magic/magic_8.3.105+ds.1.orig.tar.xz", "main magic-8.3.105/drc/DRCmain.c:284:15: style: Condition 'area!=NULL' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "The same warning (DRCmain.c:284:15, 'area!=NULL' is always true) appears in both main and your output. Only the attached notes differ, so this is a message-text or ordering difference, not a real change. The warning is a true positive either way: 'area = &scx->scx_area' can never be NULL." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/m/mgen/mgen_5.02.c+really5.1.1+dfsg.orig.tar.xz", "your mgen-5.1.1/protolib/src/common/protoLFSR.cpp:427:45: style: Condition 'NULL==t' is always false [knownConditionTrueFalse]\n mgen-5.1.1/protolib/src/common/protoLFSR.cpp:427:15: note: Assuming that condition 'NULL==c' is not redundant\n mgen-5.1.1/protolib/src/common/protoLFSR.cpp:424:15: note: c is assigned 'new UINT32[numWords]' here.\n mgen-5.1.1/protolib/src/common/protoLFSR.cpp:426:17: note: Assignment 't=new UINT32[numWords]', assigned value is symbolic=c\n mgen-5.1.1/protolib/src/common/protoLFSR.cpp:427:45: note: Condition 'NULL==t' is always false" ], "verdict": { "verdict": "neutral", "confidence": "low", "explanation": "The warning text at 427:45 ('NULL==t' is always false) is the same in main and in the PR. Only the attached note chain differs, and main's notes are not shown. The finding itself is correct: `t` comes from a plain `new UINT32[numWords]`, which throws on failure and never returns NULL, so the check is redundant. The PR's note reasoning ('assigned value is symbolic=c') is odd but reaches the same correct conclusion. Users see no meaningful change." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/m/muse/muse_4.3.0.orig.tar.gz", "main muse-sequencer-muse-8d1030d/src/libs/wave/wave.cpp:331:12: style: Condition 'finfo' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "The same knownConditionTrueFalse warning at wave.cpp:331:12 ('finfo' is always true) is still reported by the PR build. Only the attached note lines differ, so the output changed in formatting, not in substance. The warning is correct: line 331 is inside `if(finfo)` and `finfo` is not modified in between. This is still a true positive, so removing this 'main' line does not affect users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/n/nabi/nabi_1.0.1.orig.tar.gz", "main nabi-nabi-1.0.1/src/ic.c:248:13: style: Condition 'ret==-1' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "`ret` is a `size_t` that holds the result of `g_iconv`. In `ret == -1`, the `-1` is converted to SIZE_MAX, so the comparison is true exactly when `g_iconv` returns `(size_t)-1` on failure. That is the documented iconv error check. Main reported the condition as always false, most likely because it compared against a signed -1 and ignored the unsigned conversion. That was a false positive. The PR's improved implicit conversion handling removes it." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/n/nmh/nmh_1.8~RC2.orig.tar.gz", "your nmh-1.8-RC2/uip/burst.c:330:43: style: Condition 'wasdlm' is always true [knownConditionTrueFalse]\n nmh-1.8-RC2/uip/burst.c:329:36: note: Assuming that condition 'wasdlm' is not redundant\n nmh-1.8-RC2/uip/burst.c:330:43: note: Condition 'wasdlm' is always true" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "The same warning appears in both main and your. Only the attached notes differ: your output now shows the error path ('Assuming that condition wasdlm is not redundant' at line 329). The code checks `wasdlm` at line 329 (`... && wasdlm`), so inside the guarded statement `wasdlm` is indeed always true at line 330. The warning is a true positive in both versions. Adding the note is at most a slight improvement in message context and does not change what users are told." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/o/opencity/opencity_0.0.6.5stable.orig.tar.bz2", "your opencity-0.0.6.5stable/src/tinyxml/tinyxmlparser.cpp:482:9: style: Condition '!*(p+2)' is always false [knownConditionTrueFalse]\n opencity-0.0.6.5stable/src/tinyxml/tinyxmlparser.cpp:446:34: note: Assuming that condition '*(p+2)' is not redundant\n opencity-0.0.6.5stable/src/tinyxml/tinyxmlparser.cpp:482:9: note: Condition '!*(p+2)' is always false" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "Line 446 only enters the block when `*(p+2)` is non-zero. So the check `!*(p+2)` at line 482 is indeed always false, and the warning is a true positive both before and after the PR. The \"your\" version adds an error-path note pointing to the condition at line 446, which explains why the condition is always false. The warning itself is unchanged; it is now better explained, which is a small improvement." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/p/packup/packup_0.6.orig.tar.gz", "main packup-0.6/types.hh:230:25: style: Condition 'sz+(sz==0)' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "regression", "confidence": "high", "explanation": "`sz` is an `unsigned int`. In `sz+(sz==0)`, if `sz==0` the sum is 1, and otherwise it is `sz` itself, which is non-zero. The `?:` operator therefore always picks 1, so `8*(...?1:0)+1` is always 9. This is a real precedence bug: the intended code was clearly `8*(sz+(sz==0?1:0))+1`. The old warning was a true positive. The PR stops using range bounds for unsigned `+` when it cannot prove there is no wrap. Here it cannot, since `sz` has no upper bound. As a result it loses the fact that the sum is non-zero, even though this sum can never wrap to 0. Removing this true positive is a regression." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/p/polymake/polymake_4.15.orig.tar.bz2", "main polymake-4.15/lib/core/src/Rational.cc:41:28: style: Condition 'numerator_digits==0' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "`den` comes from strchr(s, '/'), so `den - s` is 0 when the string starts with '/'. That makes the check `numerator_digits == 0` reachable: it is the guard against an empty numerator. The old warning that the condition is always false was a false positive. Main probably got there by wrongly concluding that a non-null pointer minus a pointer cannot be 0. The PR changes infer and pointer-difference handling, and removing this false positive is an improvement." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/p/proftpd-mod-vroot/proftpd-mod-vroot_0.9.12.orig.tar.gz", "your proftpd-mod_vroot-0.9.12/path.c:195:11: style: Condition 'ptr==path' is always false [knownConditionTrueFalse]\n proftpd-mod_vroot-0.9.12/path.c:189:11: note: Assuming that condition 'ptr==path' is not redundant\n proftpd-mod_vroot-0.9.12/path.c:195:11: note: Condition 'ptr==path' is always false" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Both main and the PR report the same warning at path.c:195, and it is a true positive. If `ptr == path` at line 189, the function returns at 191. `ptr` is not changed before line 195, so `ptr == path` there is always false. The diff only shows the same message re-emitted; the difference is probably in the attached notes or error path, which still point correctly to line 189. Users see no meaningful change." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/p/pyparted/pyparted_3.13.0.orig.tar.gz", "your pyparted-3.13.0/src/pydevice.c:820:9: style: Condition 'ret' is always true [knownConditionTrueFalse]\n pyparted-3.13.0/src/pydevice.c:806:13: note: Assuming that condition 'ret==0' is not redundant\n pyparted-3.13.0/src/pydevice.c:820:9: note: Condition 'ret' is always true" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "Main and the PR report the same warning at the same location with the same text. Only the shown supporting notes differ, which is a presentation change. The warning is correct: after `if (ret == 0) return NULL;`, `ret` is nonzero, so `if (ret)` is always true. This pattern repeats across all the paired main/your lines for this package. The change makes no meaningful difference to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/p/pypy3/pypy3_8.0.1+dfsg.orig.tar.xz", "main pypy3.11-v8.0.1-src/pypy/module/cpyext/test/nanobind2.c:215:9: style: Condition '!wrapper' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "Line 209 returns early when `wrapper` is null, so at line 215 `!wrapper` really is always false. The warning is a true positive. The same warning is still reported by 'your'; only its text or details changed. The 'your' version adds the error path notes: 'Assuming that condition '!wrapper' is not redundant' at 209, then 'always false' at 215. This gives the user more accurate context, so the change is a small improvement in message quality." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/r/r-cran-udunits2/r-cran-udunits2_0.13.2.2.orig.tar.gz", "your udunits2/src/udunits2_R.c:157:12: style: Condition 'from==NULL' is always false [knownConditionTrueFalse]\n udunits2/src/udunits2_R.c:151:12: note: Assuming that condition 'from==NULL' is not redundant\n udunits2/src/udunits2_R.c:157:12: note: Condition 'from==NULL' is always false" ], "verdict": { "verdict": "improvement", "confidence": "low", "explanation": "Line 157 re-checks `from == NULL` right after an earlier `if (from == NULL) return;` at line 151. This is a real copy-paste bug: the check should test `to`. Both versions report the same true positive. The PR version adds an error path pointing to line 151, which explains why `from` cannot be NULL. That makes the warning easier to understand, a small improvement in message quality." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/r/rheolef/rheolef_7.2.orig.tar.gz", "main rheolef-7.2/util/bamg/Mesh2.cpp:1045:18: style: Condition 'ToSwap' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "Inside the `if (ToSwap)` block at line 1042, the ternary `ToSwap ? dets1 : detsa` at line 1045 really is always true. The warning is a true positive and stays a true positive. The PR only changed how the message is presented: it now includes notes explaining why ('Assuming that condition ToSwap is not redundant' at 1042). That makes the diagnostic more informative, so the removed 'main' line is replaced by an equivalent but better-explained 'your' line." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/s/scilab/scilab_2024.1.0+dfsg1.orig.tar.xz", "main scilab-2024.1.0/scilab/modules/scicos_blocks/src/c/shift_32_RC.c:51:31: style: Operator '|' with one operand equal to zero is redundant. [badBitmaskCheck]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "`k = (unsigned long)pow(2, 31)` is 2147483648, not 0. Main's old `castValue` set any float value outside the `int` range to 0, so it wrongly treated `k` as zero. That produced the false 'operand equal to zero' badBitmaskCheck on `y[i] | k`. The PR converts floats correctly against the destination type's range (`unsigned long`), so this false positive is gone." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/s/speechd-up/speechd-up_0.5~20110719.orig.tar.gz", "main speechd-up-0.5~20110719/speechd-up.c:358:16: style: Condition 'enc_bytes==-1' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "high", "explanation": "enc_bytes is size_t. In 'enc_bytes == -1' the -1 is converted to SIZE_MAX, which is exactly what iconv returns on error ((size_t)-1). The condition can be true, so main's 'always false' warning was a false positive. The PR's improved implicit-conversion handling of the comparison operands removes it." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/s/squeak-vm/squeak-vm_4.10.2.2614+20120917~dfsg.orig.tar.xz", "main opensmalltalk-vm-00aa355bd2f0fcfbe9b7110a70b486f89c85c55d/platforms/unix/vm-sound-pulse/sqUnixSoundPulseAudio.c:663:8: style: Condition '!audioIn.open' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Both versions report the same knownConditionTrueFalse warning at line 663. The PR only adds an error path to it.\n\nThe warning itself is unchanged. Line 635 already breaks if `!audioIn.open`. Between line 635 and line 663 only `rc` is assigned and the `isFree` check (printf + continue) runs, so single-threaded the condition is indeed always false. If another thread changes `audioIn.open`, the warning would be a false positive in both versions.\n\nThe added notes are of mixed value. The note at 658 points at the `isFree` check, which does not matter for the warning. The note at 635 does help explain it. Overall the change makes no real difference to users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/t/tclx/tclx_8.6.3+dfsg.orig.tar.xz", "main tclx-8.6.3/unix/tclXunixOS.c:577:24: style: Condition 'pid<-1' is always true [knownConditionTrueFalse]\n tclx-8.6.3/unix/tclXunixOS.c:574:24: note: Assuming that condition 'pid==-1' is not redundant\n tclx-8.6.3/unix/tclXunixOS.c:577:24: note: Condition 'pid<-1' is always true" ], "verdict": { "verdict": "regression", "confidence": "medium", "explanation": "In TclXOSkill, `pid` is a signed `pid_t`. The else-if chain tests `pid > 0`, then `pid == 0`, then `pid == -1`. When control reaches `else if (pid < -1)`, pid is \u2264 0, \u2260 0 and \u2260 -1, so the condition really is always true. The removed knownConditionTrueFalse warning was a correct, if minor, style finding. The PR's changes to infer/path handling (path splitting, dropping negative-path values, impossible-point matching) appear to have lost the combined impossible facts needed here. That is a lost true positive." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/t/termpaint/termpaint_0.3.1.orig.tar.gz", "your termpaint-0.3.1/termpaint.c:2135:50: style: Condition 'term->primary.height' is always true [knownConditionTrueFalse]\n termpaint-0.3.1/termpaint.c:2135:26: note: Assuming that condition 'term->primary.height' is not redundant\n termpaint-0.3.1/termpaint.c:2135:50: note: Condition 'term->primary.height' is always true" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "Line 2135 tests `term->primary.height && term->primary.height`, a duplicated operand. Once the first operand is true, the second is always true, so the knownConditionTrueFalse warning is a true positive in both versions. The PR keeps the warning and adds an error path: 'Assuming that condition ... is not redundant' at column 26 (the first operand), then 'always true' at column 50 (the second). These notes are accurate and explain the warning better than the bare message reported by main." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/t/twoftpd/twoftpd_1.42.orig.tar.gz", "main twoftpd-1.42/main.c:122:23: style: Comparing expression of type 'char' against value 255. Condition is always false. [compareValueOutOfTypeRangeError]" ], "verdict": { "verdict": "regression", "confidence": "medium", "explanation": "`byte` is a plain `char[1]`, and `TELNET_IAC` is 255 according to the message. On the analysed platform plain char is signed, so `*byte == 255` can never be true and the Telnet IAC handling is dead code. That is a real bug: it only works on unsigned-char targets. Main reported it because it honoured the platform's default char sign. The PR now treats plain char as having unknown signedness and only warns when the result is the same for both signed and unsigned char. Since 255 fits in unsigned char, the warning is dropped. This removes a true positive and portability-relevant warning, so it is a regression." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/v/vdk2/vdk2_2.4.0.orig.tar.gz", "main vdk-2.4.0/vdk/gtkdatabox.c:1703:8: style: Condition 'list' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "The warning at line 1703 is still reported by both versions. Only the text changed: the 'your' version adds notes to the error path (the earlier condition '!box->data', and the assignment of 'box->data' to list). The warning itself is a true positive: `box->data` is checked non-null and returns early, so `list = box->data; if (list)` is always true. The main line showing up as removed is just the message-text difference in the comparison. The extra notes are accurate and slightly helpful, but the finding is essentially unchanged." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/v/veusz/veusz_4.2.1.orig.tar.gz", "main veusz-4.2.1/src/threed/scene.cpp:263:35: style: Condition 'frag.surfaceprop!=0' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "Inside the block guarded by `frag.surfaceprop != 0` at line 259, the repeated check `frag.surfaceprop!=0` at line 263 is always true. The warning is a true positive and stays in both versions. The PR only changed the message: it now carries notes pointing back to the guarding condition at line 259 as the reason. That makes the report easier to understand, so this is a slight improvement in message accuracy and context." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/w/wmbubble/wmbubble_1.54.orig.tar.gz", "main wmbubble-1.54/bubblemon.c:1314:8: style: Condition 'memscreen_enabled' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "improvement", "confidence": "medium", "explanation": "The warning at line 1314 is unchanged in substance: 'memscreen_enabled' is already checked at line 1312, so re-checking it at 1314 is redundant, and the warning is correct. Main reports it without notes. The PR version reports the same warning but adds an error path ('Assuming that condition memscreen_enabled is not redundant' at 1312). The diff lines are the same message with and without that path. The path explains why the condition is always true, so the report is more informative." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/w/wmbubble/wmbubble_1.54.orig.tar.gz", "your wmbubble-1.54/bubblemon.c:1314:8: style: Condition 'memscreen_enabled' is always true [knownConditionTrueFalse]\n wmbubble-1.54/bubblemon.c:1312:7: note: Assuming that condition 'memscreen_enabled' is not redundant\n wmbubble-1.54/bubblemon.c:1314:8: note: Condition 'memscreen_enabled' is always true" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "Line 1314 tests 'memscreen_enabled' again inside the block guarded by 'memscreen_enabled && ...' at line 1312, so the condition really is always true. The warning is a true positive both before and after the PR. Only the main line has no notes and the your line has the same message plus notes, so the difference is most likely just in how the error path is shown. It does not change what users are warned about." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/w/wvstreams/wvstreams_4.6.1.orig.tar.gz", "main wvstreams-4.6.1/urlget/wvhttppool.cc:227:13: style: Condition '!i->outstream' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "medium", "explanation": "The same knownConditionTrueFalse warning is still reported at line 227. Only its error path changed: the PR version adds a note pointing to the earlier `!i->outstream` check at line 211. The warning itself is questionable, because `unconnect(s)` may mark the URL bad and clear `outstream`, as the source comment says. But that is true for both main and the PR, so for users the change is cosmetic." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/x/xfireworks/xfireworks_1.3.orig.tar.gz", "main xfireworks-1.3.orig/Piece.c:317:9: style: Condition 'w1' is always true [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "The PR did not remove or add a warning here. The 'main' line and the 'your' line are the same warning (Piece.c:317 knownConditionTrueFalse 'w1' is always true). The diff only shows it as changed because of note lines or formatting. The warning is correct in both versions: w1 is non-NULL inside the loop because of the while condition on line 296. So nothing changes for users." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/x/xfireworks/xfireworks_1.3.orig.tar.gz", "your xfireworks-1.3.orig/Piece.c:317:9: style: Condition 'w1' is always true [knownConditionTrueFalse]\n xfireworks-1.3.orig/Piece.c:296:34: note: Assuming that condition '(w1=GetWord2(stream))!=NULL' is not redundant\n xfireworks-1.3.orig/Piece.c:317:9: note: Condition 'w1' is always true" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "The warning at Piece.c:317 is reported by both main and your branch. Only the attached notes differ, so the diff shows it as a removed and an added line. The finding is correct: w1 is non-NULL inside the while loop, so `if (w1)` is always true. The note now points to the loop condition at line 296 as the source. Nothing changes for users in a meaningful way." } }, { "key": [ "https://ftp.debian.org/debian/pool/main/z/zynaddsubfx/zynaddsubfx_3.0.6.orig.tar.bz2", "main zynaddsubfx-3.0.6/DPF/dgl/src/nanovg/nanovg.c:1799:21: style: Condition 'lineCap==NVG_BUTT' is always false [knownConditionTrueFalse]" ], "verdict": { "verdict": "neutral", "confidence": "high", "explanation": "The same warning at line 1799 appears in both main and your output. It is listed as one 'main' line and one 'your' line only because the attached notes differ. The warning is a true positive: in the else-if branch, 'lineCap==NVG_BUTT' was already excluded by the previous 'if (lineCap == NVG_BUTT)'. Its text and position are unchanged, so nothing changes for users." } } ], "started": "2026-10-30 13:16:58 UTC", "runner": "server" }