# Blob hash of binary files in patches generated by git format patch show in full form instead of short form 4 messages from 2021-03-21 to 2021-03-22. Participants: Bagas Sanjaya, Junio C Hamano. Thread: https://gitlist.dev/t/55362 ## Bagas Sanjaya, 2021-03-21 13:05 Subject: Blob hash of binary files in patches generated by git format patch show in full form instead of short form Message-ID: <499c9922-eb42-c2a8-b4b4-8e5197ea0fc6@gmail.com> URL: https://gitlist.dev/e/499c9922-eb42-c2a8-b4b4-8e5197ea0fc6%40gmail.com ``` Thank you for filling out a Git bug report! Please answer the following questions to help us understand your issue. What did you do before the bug happened? (Steps to reproduce your issue) I'm trying to do format-patch with binary files (images) in commits. In each commit, it adds an image and its alt description text in separated file. Full steps: First, initialize empty repo and populate it with commits: - cd /tmp - mkdir bin-patch && cd bin-patch - git init - echo "test format-patch binary files" > README - git add * && git commit -m "init README" - git checkout -b test - wget -c [1] -O stackoverflow.png && echo "Stack Overflow" > stackoverflow.alt - git add * && git commit -m "Add Stack Overflow logo" - wget -c [2] -O idntm.jpg && echo "Indonesia's Next Top Model cast" > idntm.alt - git add * && git commit -m "Add IdNTM cast poster" Now prepare patches as usual (with cover letter ignored for this purpose): - git format-patch --cover-letter -M master (image link): [1]: https://upload.wikimedia.org/wikipedia/commons/thumb/0/02/Stack_Overflow_logo.svg/1280px-Stack_Overflow_logo.svg.png [2]: https://upload.wikimedia.org/wikipedia/en/9/9f/IndonesiaNTM1Cast.jpg What did you expect to happen? (Expected behavior) Blob hash in the `index` header of generated patches for binary file (image) use short form (7 characters), just like for text file (alt description). What happened instead? (Actual behavior) Blob hash in the `index` stanza of generated patches for image use full (long) form. For first patch (Stack Overflow commit), diff header for the image read: ˋˋˋ diff --git a/stackoverflow.png b/stackoverflow.png new file mode 100644 index 0000000000000000000000000000000000000000..969908ad3161a66af31f2441cfea4ae002a5ec67 ˋˋˋ while diff header for alt description read: ˋˋˋ diff --git a/stackoverflow.alt b/stackoverflow.alt new file mode 100644 index 0000000..9368417 ˋˋˋ Similarly, for the second patch (INTM poster), diff header for the image read: ˋˋˋ diff --git a/idntm.jpg b/idntm.jpg new file mode 100644 index 0000000000000000000000000000000000000000..8921ab9540e0a36a53f8c6632482fb04d5d0cc6c ˋˋˋ while diff header for alt description read: ˋˋˋ diff --git a/idntm.alt b/idntm.alt new file mode 100644 index 0000000..719feb9 ˋˋˋ What's different between what you expected and what actually happened? Blob hash for binary files are shown in full form, as opposed to blob hash for text files. Anything else you want to add: (none) Please review the rest of the bug report below. You can delete any lines you don't wish to share. [System Info] git version: git version 2.31.0.29.g98164e9585 cpu: x86_64 built from commit: 98164e9585e02e31dcf1377a553efe076c15f8c6 sizeof-long: 8 sizeof-size_t: 8 shell-path: /bin/sh uname: Linux 5.11.6-kernelorg-upstream-generic #1 SMP Fri Mar 12 06:35:27 WIB 2021 x86_64 compiler info: gnuc: 9.3 libc info: glibc: 2.31 $SHELL (typically, interactive shell): /bin/bash [Enabled Hooks] (none) -- An old man doll... just what I always wanted! - Clara ``` ## Junio C Hamano, 2021-03-21 17:31 Subject: Re: Blob hash of binary files in patches generated by git format patch show in full form instead of short form Message-ID: URL: https://gitlist.dev/e/xmqqblbcbehd.fsf%40gitster.g In-Reply-To: <499c9922-eb42-c2a8-b4b4-8e5197ea0fc6@gmail.com> ``` Bagas Sanjaya writes: > What's different between what you expected and what actually happened? > > Blob hash for binary files are shown in full form, as opposed to blob hash > for text files. This is working as intended, designed and implemented. The textual patch is meant to be applicable on target text that may even have been slightly modified from the original from which the patch was taken, and the abbreviated object name on the "index" line is there mostly for human's sanity check and as a visual aid. Ordinarily it is not used to actually find the matching blob object (and it is not an error if there is no matching blob object in the repository that a patch application is attempted in). But the binary patch is designed to be applicable only to an exact copy of the original and nowhere else. The object name is given in full, instead of using abbreviated form, to ensure that we do not try to apply a binary patch to an object whose name is "similar". Thanks. ``` ## Bagas Sanjaya, 2021-03-22 05:47 Subject: Re: Blob hash of binary files in patches generated by git format patch show in full form instead of short form Message-ID: <56cde808-95c3-2e22-2dab-880061d51473@gmail.com> URL: https://gitlist.dev/e/56cde808-95c3-2e22-2dab-880061d51473%40gmail.com In-Reply-To: ``` On 22/03/21 00.31, Junio C Hamano wrote: > Bagas Sanjaya writes: > >> What's different between what you expected and what actually happened? >> >> Blob hash for binary files are shown in full form, as opposed to blob hash >> for text files. > > This is working as intended, designed and implemented. > > The textual patch is meant to be applicable on target text that may > even have been slightly modified from the original from which the > patch was taken, and the abbreviated object name on the "index" line > is there mostly for human's sanity check and as a visual aid. > Ordinarily it is not used to actually find the matching blob object > (and it is not an error if there is no matching blob object in the > repository that a patch application is attempted in). > > But the binary patch is designed to be applicable only to an exact > copy of the original and nowhere else. The object name is given in > full, instead of using abbreviated form, to ensure that we do not > try to apply a binary patch to an object whose name is "similar". > > Thanks. > Hmm... but I don't see that in the documentation for git format-patch. Maybe I need to send doc update. -- An old man doll... just what I always wanted! - Clara ``` ## Bagas Sanjaya, 2021-03-22 10:06 Subject: Re: Blob hash of binary files in patches generated by git format patch show in full form instead of short form Message-ID: URL: https://gitlist.dev/e/a29052c9-b309-080a-537e-af6114116930%40gmail.com In-Reply-To: <56cde808-95c3-2e22-2dab-880061d51473@gmail.com> ``` Oh oh oh, I see git diff format for generating patches in documentation for git diff-files. On 22/03/21 12.47, Bagas Sanjaya wrote: > On 22/03/21 00.31, Junio C Hamano wrote: >> Bagas Sanjaya writes: >> >>> What's different between what you expected and what actually happened? >>> >>> Blob hash for binary files are shown in full form, as opposed to blob hash >>> for text files. >> >> This is working as intended, designed and implemented. >> >> The textual patch is meant to be applicable on target text that may >> even have been slightly modified from the original from which the >> patch was taken, and the abbreviated object name on the "index" line >> is there mostly for human's sanity check and as a visual aid. >> Ordinarily it is not used to actually find the matching blob object >> (and it is not an error if there is no matching blob object in the >> repository that a patch application is attempted in). >> >> But the binary patch is designed to be applicable only to an exact >> copy of the original and nowhere else.  The object name is given in >> full, instead of using abbreviated form, to ensure that we do not >> try to apply a binary patch to an object whose name is "similar". >> >> Thanks. >> > > Hmm... but I don't see that in the documentation for git format-patch. > Maybe I need to send doc update. > -- An old man doll... just what I always wanted! - Clara ```