Base Media File Format #1475

1div0 · 2021-02-16T09:27:06Z

Forgot to create PR against 0.27-maintenance instead of master. Sorry for the noise.

… isobmff code.

…. Removed C++11 style code. Removed unused code.

…INGS_AS_ERRORS=On

Reason: clang/ubuntu fails saying "Package cmake is not available, but is referred to by another package.".

lgtm-com · 2021-02-16T09:40:09Z

This pull request introduces 1 alert when merging f0a321d into 443944d - view on LGTM.com

new alerts:

1 for Empty branch of conditional

kmilos · 2021-02-16T10:50:38Z

src/bmffimage.cpp

+            case TAG_avif:
+                return "image/avif";
+
+            case TAG_heic:


Seems there is dedicated image/heic mime type: https://www.iana.org/assignments/media-types/media-types.xhtml#image

kmilos · 2021-02-16T10:52:44Z

src/bmffimage.cpp

+
+#define TAG_ftyp ID("ftyp") /**< File type box */
+#define TAG_avif ID("avif") /**< AVIF */
+#define TAG_heic ID("heic") /**< HEIF */


Typo: HEIF->HEIC

These are different tags for files already produced. So no typo, even if it seems obvious.

heic means container
heif means format
heix is extended precision, i.e 10 bits per channel, however I haven't seen such files yet.

The Canon HEIV files that Gordon sent have brand: heix. I'm hoping to setup the test environment tomorrow (2021-02-26). I don't want to drop big fat files into test/data in the repos. The test files will be stored somewhere in GitHub and downloaded when cmake -DEXIV2_ENABLE_BMFF=On is used.

Bingo! I was waiting for a first encounter with 10 bpc files.

lgtm-com · 2021-02-16T11:01:37Z

This pull request introduces 1 alert when merging 7b5854e into 443944d - view on LGTM.com

new alerts:

1 for Empty branch of conditional

lgtm-com · 2021-02-16T11:24:36Z

This pull request introduces 1 alert when merging 81e0f99 into 443944d - view on LGTM.com

new alerts:

1 for Empty branch of conditional

https://ci.appveyor.com/project/piponazo/exiv2-wutfp/builds/37795261/job/23l6a0u0wt6ax79l

lgtm-com · 2021-02-16T12:49:41Z

This pull request introduces 1 alert when merging 4fa0a88 into 443944d - view on LGTM.com

new alerts:

1 for Empty branch of conditional

lgtm-com · 2021-02-23T20:22:13Z

This pull request introduces 5 alerts when merging 6d13e44 into c6de72f - view on LGTM.com

new alerts:

3 for Wrong type of arguments to formatting function
2 for Expression has no effect

lgtm-com · 2021-02-23T21:10:52Z

This pull request introduces 5 alerts when merging 630fb23 into c6de72f - view on LGTM.com

new alerts:

3 for Wrong type of arguments to formatting function
2 for Expression has no effect

README.md

include/exiv2/exiv2.hpp

man/man1/exiv2.1

kmilos · 2021-03-08T10:28:58Z

README.md

@@ -41,7 +41,7 @@ The file ReadMe.txt in a build bundle describes how to install the library on th
   16. [Cross Platform Build and Test on Linux for MinGW](#2-16)
   17. [Building with C++11 and other compilers](#2-17)
   18. [Static and Shared Libraries](#2-18)
-   19. [Support for bmff files (CR3, HIF, AVIF and HEIC)](#2-19)
+   19. [Support for bmff files (CR3, HEIF and AVIF)](#2-19)


Hm, I didn't mean remove HEIC (and from main page neither below)... As per Wikipedia, I believe we should have HEIF, HEIC, and AVIF documented.

The HIF/HEIF is the same as JPG/JPEG IMHO, no need to handle specially.

@clanmills Maybe I just do a PR with what I had in mind?

Yes. My telepathy is down at the moment! Can you wait until this PR has been merged into 0.27-maintenance and then open a new PR.

Sorry, saw this late and already pushed - just minor formatting changes though...

Milos. Please don't make cosmetic changes to existing code that isn't part of the PR. It introduces changes between old-master and 0.27-maintenance which makes porting fixes from old-master more tricky. @neheb did a lot of work last Fall on 'master' and it would be good to see if we can merge his great work into 0.27-maintenance.

@neheb: I'm sorry to let you know that the Team has decided to hibernate 'master'. #1466 (comment)

@piponazo and @hassec have both proposed reformatting all the code with clang-tidy. I think we should consider doing that on 'main' when that ships as 0.28. Currently scheduled for 2021-09-15.

Apologies Robin, I reverted my commits now. I'll do just the relevant subset once this PR is merged.

@kmilos Thank you for both making the change and reverting the change. I know you intended well with those changes.

clanmills · 2021-03-09T10:44:57Z

C'mon guys. Let's get this reviewed, approved and merged. Please log any testing you have done other than the test suite.

There are 168 CR3 files (in 4.5GB) on raw.pixls.us. 26mb/image. Exiv2 reads them all in 12 seconds (on a 7 year old mac mini). 70msecs/image. It's 90% overhead (waiting for the disk?). So, the library reads > 100 images/second. Good, eh?

528 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $ type finder
finder is aliased to `find . -depth -iname'
529 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $ finder "*.cr3" -print0 | xargs -0 ls -l | wc
    168    1950   18134
530 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $ c=0;for i in $(finder "*.cr3" -print0 | xargs -0 ls -l | cut -d' ' -f 7); do c=$((c+i)); done; echo $(( c/(1024*1024) ))gb
4750gb
531 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $ finder "*.cr3" -print0 | xargs -0 exiv2 -K Exif.Image.DateTime 
./Pixls/Canon/Canon PowerShot G5 X Mark II/Canon_-_PowerShot_G5_X_Mark_II_-_RAW_(3:2).CR3  Exif.Image.DateTime                          Ascii      20  2019:10:27 10:00:47
./Pixls/Canon/Canon PowerShot G5 X Mark II/Canon_-_PowerShot_G5_X_Mark_II_-_CRAW_(1:1).CR3  Exif.Image.DateTime                          Ascii      20  2019:10:27 10:05:09
./Pixls/Canon/Canon PowerShot G5 X Mark II/Canon_-_PowerShot_G5_X_Mark_II_-_CRAW_(4:3).CR3  Exif.Image.DateTime                          Ascii      20  2019:10:27 10:02:12
...
532 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $ time finder "*.cr3" -print0 | xargs -0 exiv2 -K Exif.Image.DateTime | wc
    168    1446   23174

real	0m12.734s
user	0m1.023s
sys	0m2.582s
533 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $

The tvisitor code from my book takes 11 seconds. That's proof that Exiv2 is as efficient as tvisitor. tvisitor is very streamlined. I don't think it's possible to read those files faster.

535 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $ finder "*.cr3" -print0 | xargs -0 tvisitor -pR | grep Exif.Image.DateTime | head -3
           130 | 0x0132 Exif.Image.DateTime              |     ASCII |       20 |       230 | 2019:10:27 10:00:47
           130 | 0x0132 Exif.Image.DateTime              |     ASCII |       20 |       230 | 2019:10:27 10:05:09
           130 | 0x0132 Exif.Image.DateTime              |     ASCII |       20 |       230 | 2019:10:27 10:02:12
536 rmills@rmillsmm-local:/Users/Shared/Jenkins/Home/userContent/testfiles $

piponazo

I did not put too much attention in the implementation details, but overall the job you guys have done here looks great! 👍

piponazo · 2021-03-09T12:04:45Z

include/exiv2/bmffimage.hpp

+// namespace extensions
+namespace Exiv2
+{
+    EXIV2API bool enableBMFF(bool enable = true);


Should not this function has return type void ? I see that in the examples and documentation we do not use the returned value.

The return value is important. It tells the caller if the code has been built with BMFF support. In the documentation and samples, we ignore the return value. However, "real" applications can use this to "gray out" a UI option such as "Use bmff support", or to post an alert such as bmff support cannot be enabled on this platform.

piponazo · 2021-03-09T12:07:52Z

include/exiv2/jp2image.hpp

@@ -42,7 +42,7 @@ namespace Exiv2
    // Add JPEG-2000 to the supported image formats
    namespace ImageType
    {
-        const int jp2 = 15;                     //!< JPEG-2000 image type
+        const int jp2 = 19;                     //!< JPEG-2000 image type


Why is the new type using the value 15 and jp2 value has been updated from 15 to 19? Was not possible to keep the previous values for jp2 and use a new one for BMFF?

15 is used by bmff, and 19 by jp2.

Let's replace that with an enum in 'main'. I didn't want change it to an enum for a "dot" because if it's not broken, don't fix it. However, when we get to 'main/0.28' we can fix this.

509 rmills@rmillsm1:~/gnu/github/exiv2/bmff $ grep 'const int' include/exiv2/*image.hpp include/exiv2/bmffimage.hpp: const int bmff = 15; //!< BMFF (bmff) image type (see class BMFF) include/exiv2/bmpimage.hpp: const int bmp = 14; //!< Windows bitmap (bmp) image type (see class BmpImage) include/exiv2/cr2image.hpp: const int cr2 = 7; //!< CR2 image type (see class Cr2Image) include/exiv2/crwimage.hpp: const int crw = 3; //!< CRW image type (see class CrwImage) include/exiv2/epsimage.hpp: const int eps = 18; //!< EPS image type include/exiv2/gifimage.hpp: const int gif = 11; //!< GIF image type (see class GifImage) include/exiv2/image.hpp: const int none = 0; //!< Not an image include/exiv2/jp2image.hpp: const int jp2 = 19; //!< JPEG-2000 image type include/exiv2/jpgimage.hpp: const int jpeg = 1; //!< JPEG image type (see class JpegImage) include/exiv2/jpgimage.hpp: const int exv = 2; //!< EXV image type (see class ExvImage) include/exiv2/mrwimage.hpp: const int mrw = 5; //!< MRW image type (see class MrwImage) include/exiv2/orfimage.hpp: const int orf = 9; //!< ORF image type (see class OrfImage) include/exiv2/pgfimage.hpp: const int pgf = 17; //!< PGF image type (see class PgfImage) include/exiv2/pngimage.hpp: const int png = 6; //!< PNG image type (see class PngImage) include/exiv2/psdimage.hpp: const int psd = 12; //!< Photoshop (PSD) image type (see class PsdImage) include/exiv2/rafimage.hpp: const int raf = 8; //!< RAF image type (see class RafImage) include/exiv2/rw2image.hpp: const int rw2 = 16; //!< RW2 image type (see class Rw2Image) include/exiv2/tgaimage.hpp: const int tga = 13; //!< Truevision TARGA (tga) image type (see class TgaImage) include/exiv2/tiffimage.hpp: const int tiff = 4; //!< TIFF image type (see class TiffImage) include/exiv2/tiffimage.hpp: const int dng = 4; //!< DNG image type (see class TiffImage) include/exiv2/tiffimage.hpp: const int nef = 4; //!< NEF image type (see class TiffImage) include/exiv2/tiffimage.hpp: const int pef = 4; //!< PEF image type (see class TiffImage) include/exiv2/tiffimage.hpp: const int arw = 4; //!< ARW image type (see class TiffImage) include/exiv2/tiffimage.hpp: const int sr2 = 4; //!< SR2 image type (see class TiffImage) include/exiv2/tiffimage.hpp: const int srw = 4; //!< SRW image type (see class TiffImage) include/exiv2/webpimage.hpp: const int webp = 23; //!< Treating webp as an image type> 510 rmills@rmillsm1:~/gnu/github/exiv2/bmff $

I've realised during lunch that you're asking "why not leave jp2 unchanged and set bmff=19?". When I started working with Peter, the only way I could change anything was to send him a patch and that's what I did. I didn't want to discuss what that code means, so I accepted his 15 (even though it's wrong).

You're right. Now that we're cleaning up and the end, I should restore jp2 = 15 and set bmff = 19. Good Catch.

piponazo · 2021-03-09T12:15:12Z

src/bmffimage.cpp

+        return result;
+    }
+
+    long BmffImage::boxHandler(std::ostream& out /* = std::cout*/ , Exiv2::PrintStructureOption option /* = kpsNone */,int depth /* =0 */)


I guess the comments can be removed from here, right? 😇

Comments? Are you asking me to add comments? I've put comments in the declaration in bmffimage.hpp

I am not asking you to add comments. It looks like the IDE you were using added those default values:

/* = std::cout*/ /* = kpsNone */

from the .h and commented them out in the .cpp. If possible, it would be good to remove them from the .cpp.

Ah, right. It's a common convention to repeat the defaults in the cpp. Of course that have to be in comments or the compiler gets upset.

I'll change that. NP.

LeoHsiao1 · 2021-03-09T12:23:03Z

@clanmills
I can execute your new test case. But I'm simplifying it a little bit.

LeoHsiao1

This PR adds three test cases: tests/bugfixes/github/test_pr1475_*.py
And two functions, runTest(), verbose_version(), have been added to tests/system_tests.py without affecting the existing test code.
I've passed the test cases:

test_run (test_pr1475_AVIF.pr_1475_avif_avif) ... ok
test_run (test_pr1475_AVIF.pr_1475_exif_xmp_avif) ... ok
test_run (test_pr1475_AVIF.pr_1475_metadata2_avif) ... ok
test_run (test_pr1475_HEIC.pr_1475_2021_heic) ... ok
test_run (test_pr1475_HEIC.pr_1475_IMG_3578_heic) ... ok
test_run (test_pr1475_HEIC.pr_1475_Stonehenge_heic) ... ok
test_run (test_pr1475_HEIC.pr_1475_heic_heic) ... ok
test_run (test_pr1475_HIF.pr_1475_Canon_hif) ... ok
test_run (test_pr1475_HIF.pr_1475_Sony_hif) ... ok

clanmills · 2021-03-09T13:11:27Z

@LeoHsiao1 Thanks for looking at this. I'll look at your changes later today.

BTW. I was working on the release notes for v0.27.4 this morning and of course you get acknowledged. By the end of this week, I hope to have removed the legacy bash tests and your code with have the responsibility for performing this important test activity. Make and bash will be no longer required to test msvc builds. #1018 (comment).

clanmills · 2021-03-09T14:09:00Z

@LeoHsiao1 I can see you've simplified the python code. Thanks. I'm not sure what is going on in git. I don't see a difference anywhere. The important thing is that the simpler/modified code works.

I've thought of a possible bug. The environment string EXIV2_BINDIR can be used to specify the location of the exiv2. My code to modify PATH and LD_LIBRARY_PATH doesn't use EXIV2_BINDIR. However, the code could be correct because Dan's file tests/suite.conf locates exiv2. Could you investigate this and let me know. Thanks.

…quested.

hassec · 2021-03-09T14:21:11Z

src/futils.cpp

+        size_t input_length = in ? ::strlen(in) : 0;
+        if (!in || input_length % 4 != 0) return result;
+
+        unsigned char encoding_table[]= {'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H',


I think it would be nice if w can move this table out of this function and share it with the encode function, no?

Yes. I thought of that. In decode it's a char** and in encode it's unsigned char** (or the other way). And I thought "what happens if this is exposed as an external in the library?". So, I thought leave it alone - it's working.

Since we are in a .cpp we can just put it into an anonymous namespace and wouldn't have to worry about leaking it.

That's probably right. I'm worrying about nothing. Tomorrow I'll move encoding_table into a static outside the encode and decode functions.

src/actions.cpp

clanmills · 2021-03-09T15:44:55Z

@LeoHsiao1. Can you put my code into your file tests/bash_tests/utils.py. Then we can leave Dan's code untouched by the PR. My code looks so different from Dan's. Let's aim to leave Dan's code untouched.

I've thought of several things which we need concerning test for v0.27.4. Can you deal with the following and submit a separate PR as the following has nothing to do with bmff support.

We have to rewrite test/version_test.sh and test/unit_test.sh in python. I think the code system_tests.verbose_version(True) will do what we need. We need a little python magic to cause the following to work without bash:

$ make unit_test     or $ cmake --build . --target unit_test     # runs bin/unit_tests
$ make bash_tests    or $ cmake --build . --target bash_tests    # runs your code in tests/bash_tests
$ make python_tests  or $ cmake --build . --target python_tests  # runs Dan's code in tests/bugfixes
$ make version_test  or $ cmake --build . --target version_test  # runs and formats output of exiv2 --verbose --version 
$ make tests         or $ cmake --build . --target tests         # runs every test

And I'l like to rewire $ make bash_tests so that it doesn't use bash. It simply runs you tests/bash_tests magic.

I don't need this done today. However if this could be done at the weekend, it'll save me some time. And I really want you to totally understand and own the test code in Exiv2. You can do it.

…xiv2::enableBMFF().

…oth Exiv2::base64encode() and Exiv2::base64decode(). I'm a little uneasy about changes to the API of the shared object/DLL. I will investigate Exiv2#890 during the release process. So I have a plan to identify this if it's an issue. I think it's OK. ``` 703 rmills@rmillsmm-local:~/gnu/github/exiv2/bmff/build $ nm --demangle lib/libexiv2.dylib | grep base64 00000000000384a0 T Exiv2::base64decode(char const*, char*, unsigned long) 0000000000038340 T Exiv2::base64encode(void const*, unsigned long, char*, unsigned long) 00000000001c33a0 s Exiv2::base64_encode 704 rmills@rmillsmm-local:~/gnu/github/exiv2/bmff/build $ nm -g --demangle lib/libexiv2.dylib | grep base64 00000000000384a0 T Exiv2::base64decode(char const*, char*, unsigned long) 0000000000038340 T Exiv2::base64encode(void const*, unsigned long, char*, unsigned long) 705 rmills@rmillsmm-local:~/gnu/github/exiv2/bmff/build $ ```

…tem_tests.BT

clanmills

This is it. Time to merge. Thank you to @alexvanderberkel @hassec @piponazo @LeoHsiao1 @kmilos and @1div0. Great Team Work by Everybody.

clanmills · 2021-03-10T10:42:18Z

Shit. There's something horribly wrong and clearly caused by a collision with #1482. Git strikes again!

[ 41%] Building CXX object unitTests/CMakeFiles/unit_tests.dir/test_LangAltValueRead.cpp.o
/Users/rmills/gnu/github/exiv2/0.27-maintenance/unitTests/test_LangAltValueRead.cpp:21:35: error: no member named 'kerInvalidLangAltValue' in namespace 'Exiv2'
        ASSERT_EQ(e.code(),Exiv2::kerInvalidLangAltValue);

I'll investigate and fix it this morning.

clanmills · 2021-03-10T11:03:35Z

Storm in a tea-cup. Deleted the build directory and generated a new environment. Everything appears to be fine on macOS. I'll use the build server to build/test every platform (including UNIX) this morning.

1div0 and others added 9 commits January 24, 2021 15:09

[WIP] Add ISO/IEC Base Media File Format

92469f4

Fix image handler to give jp2image code higher priority than the next…

460a802

… isobmff code.

Rename class ISOBMFF => class bmffImage to match other image handlers…

0844e1b

…. Removed C++11 style code. Removed unused code.

Fixing Linux build/test issues.

fa39234

Fix Linux build-breaker when ENABLE_ISOBMFF=False and EXIV2_TEAM_WARN…

f5c6e75

…INGS_AS_ERRORS=On

Small corrections

e19fca6

Modified ci/install.sh to install cmake before dependencies.

2104665

Reason: clang/ubuntu fails saying "Package cmake is not available, but is referred to by another package.".

[WIP] Added box types

c608148

[WIP] Base Media File Format

f0a321d

[WIP] Redefine tags

7b5854e

kmilos reviewed Feb 16, 2021

View reviewed changes

[WIP] Another try

81e0f99

[WIP] Correction

4fa0a88

https://ci.appveyor.com/project/piponazo/exiv2-wutfp/builds/37795261/job/23l6a0u0wt6ax79l

1div0 and others added 4 commits February 22, 2021 11:01

[WIP] 64-bit length

04481dd

[WIP] Correction to make Travis CI happy

a75ac74

[WIP] Yet another type cast correction to make Travis CI happy

0360a7d

WIP: Refactored readMetadata() into recursive boxHandler()

6d13e44

Fixing a build breaker.

630fb23

clanmills added 2 commits February 24, 2021 11:55

Fix linux/CI build breaker.

ea96814

Fix msvc/CI build breakers.

0b4b7c6

clanmills assigned 1div0 Feb 24, 2021

clanmills requested review from clanmills and alexvanderberkel February 24, 2021 12:04

kmilos reviewed Mar 8, 2021

View reviewed changes

README.md Outdated Show resolved Hide resolved

README.md Outdated Show resolved Hide resolved

include/exiv2/exiv2.hpp Outdated Show resolved Hide resolved

man/man1/exiv2.1 Outdated Show resolved Hide resolved

Updated to adopt Review suggestions by @kmilos. Thank You, Milos.

ae66869

kmilos reviewed Mar 8, 2021

View reviewed changes

kmilos and others added 4 commits March 8, 2021 11:41

Include HEIC type in docs

2c0f3c2

Minor whitespace formatting

e09d093

Fix msvc build breaker.

4c52861

Revert style changes

0da1a50

Simplify functions: runTest(), verbose_version()

9d28943

piponazo reviewed Mar 9, 2021

View reviewed changes

clanmills mentioned this pull request Mar 9, 2021

Exiv2 RoadMap #1018

Open

LeoHsiao1 approved these changes Mar 9, 2021

View reviewed changes

Thank You @piponazo for the code review. I've made the changes you re…

e3fd0c6

…quested.

hassec reviewed Mar 9, 2021

View reviewed changes

src/actions.cpp Show resolved Hide resolved

hassec approved these changes Mar 9, 2021

View reviewed changes

piponazo approved these changes Mar 9, 2021

View reviewed changes

clanmills added 3 commits March 10, 2021 08:59

Following review by @piponazo, I am clarifying the bool return from E…

212f1b5

…xiv2::enableBMFF().

Move system_tests.runTest() and system_tests.verbose_version() to sys…

19aab43

…tem_tests.BT

clanmills approved these changes Mar 10, 2021

View reviewed changes

clanmills merged commit 9e2ac47 into Exiv2:0.27-maintenance Mar 10, 2021

clanmills mentioned this pull request Mar 10, 2021

Add support for ISOBMFF Files (AVIF, HEIF, CR3) #1229

Closed

clanmills mentioned this pull request Mar 22, 2021

HEIF files with 10 bits per channel are not properly decoded strukturag/libheif#468

Closed

Base Media File Format #1475

Base Media File Format #1475

Conversation

1div0 commented Feb 16, 2021

lgtm-com bot commented Feb 16, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

lgtm-com bot commented Feb 16, 2021

lgtm-com bot commented Feb 16, 2021

lgtm-com bot commented Feb 16, 2021

lgtm-com bot commented Feb 23, 2021

lgtm-com bot commented Feb 23, 2021

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

clanmills Mar 8, 2021 • edited

Choose a reason for hiding this comment

kmilos Mar 8, 2021 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

clanmills commented Mar 9, 2021

piponazo left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

LeoHsiao1 commented Mar 9, 2021

LeoHsiao1 left a comment

Choose a reason for hiding this comment

clanmills commented Mar 9, 2021

clanmills commented Mar 9, 2021 • edited

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

clanmills commented Mar 9, 2021 • edited

clanmills left a comment

Choose a reason for hiding this comment

clanmills commented Mar 10, 2021

clanmills commented Mar 10, 2021

clanmills Mar 8, 2021 •

edited

kmilos Mar 8, 2021 •

edited

clanmills commented Mar 9, 2021 •

edited

clanmills commented Mar 9, 2021 •

edited