| Patch | Description | Author | Forwarded | Bugs | Origin | Last update |
|---|---|---|---|---|---|---|
| backport/performance/19470ad786d534ba78025ae88b82013820721c26.patch | [PATCH] ReadBarcode: help gcc's auto-vectorizer with the RGB->Lum conversion This makes the `RGBToLum` loop 2-3 times faster with gcc but does currently not help llvm. No idea what msvc is able to make of this. For very specific use cases, like `ZXingReader samples/falsepositives-1/09.png -s -format qrcode -nos -bin global` this can make a substantial difference (.43ms instead of .76ms). |
axxel <awagger@gmail.com> | no | 2024-02-08 | ||
| 0001-DecodeHints-limit-the-2.2.1-c-ABI-fix-to-shared-libr.patch | DecodeHints: limit the 2.2.1 c++ ABI fix to shared library builds This should fix the problem with building libreoffice with enabled precompiled headers. See #685. |
axxel <awagger@gmail.com> | not-needed | upstream | 2023-12-13 | |
| backport/ac88bce74311fdbf376faf0bc88e90d0990d048a.patch | [PATCH] QRCode: skip extra version check The version bits have already been parsed during detection. If they would have been wrong then, we would not have ended up here. If we did, there is no point in reading them again. This fixes #688. |
axxel <awagger@gmail.com> | no | 2023-12-15 | ||
| backport/e607eb72b70e3c774821213184e1f546964db593.patch | [PATCH] README: add note about minimum c++ version (fix #689) | axxel <awagger@gmail.com> | no | 2023-12-19 | ||
| backport/handling/82be565c80495dd8c5c186a26eaeb95e3b2a0513.patch | [PATCH] Error: add check for empty ImageView in ReadBardcodes | axxel <awagger@gmail.com> | no | 2023-12-29 | ||
| backport/handling/d6177733db1e5af6b010de08079476dc045af76f.patch | [PATCH] ImageView: add overflow handling for left/top cropping params | axxel <awagger@gmail.com> | no | 2024-01-11 | ||
| backport/handling/89cac928c62c3fe833504dcef2e7f1af53b0786b.patch | [PATCH] ReadBarcode: add check for downscaleFactor | axxel <awagger@gmail.com> | no | 2024-02-06 | ||
| backport/crashfix/d1c34452b0a0a9a4936666e969c88f774f399c78.patch | [PATCH] QRCode: fix crash reported in #700 | axxel <awagger@gmail.com> | no | 2024-01-12 | ||
| backport/crashfix/6ab17fdce1b8f0e605c1eb0266da63eca261529a.patch | [PATCH] ODCode128Writer: make sure a std::exception::what() string is valid UTF8 This should fix the root cause of the iOS crash discussed in the #717. |
axxel <awagger@gmail.com> | no | 2024-02-08 | ||
| backport/cmdline/c580d1e0c0c53a58cea1789b5a9dd8e78b5a53e1.patch | [PATCH] ZXingReader: add `-binarizer <local|global|fixed>` command line option | axxel <awagger@gmail.com> | no | 2024-02-02 | ||
| backport/cmdline/e01f2d4f8f5550f63f25e236be7058327895ecc7.patch | [PATCH] ZXingReader: add `-single` option to `setMaxNumberOfSymbols(1)` | axxel <awagger@gmail.com> | no | 2024-02-07 | ||
| backport/performance/fb617df67f497e5da13c8cad7e5d5e71bcb36a3f.patch | [PATCH] cmake: drop the `-ffloat-store` handling I can't reproduce the issues that this setting was fixing when originally introduced. And here comes the kicker: removing it make the g++ Release build on my Core i9 machine 20% faster!?! (Insert eye popping emoji here.) The decode time for the full ReaderTest goes down from 5.2s to 4.2s. While the clang++ build takes 4.7s either way (does not support that setting at all). |
axxel <awagger@gmail.com> | no | 2024-02-07 |