Test-drive Tour 6

[damithc]

Similar to #154
Tour available at https://git-mastery.github.io/lessons/trail/branchingLocally/

[jovnc]

Test Summary (Hands-on + Content)

T6L1

Misc

From the graph, it appears that the repo’s working directory reflects the code at commit c instead of b.

The text for [5] seems to be referring to [6], and the text for [6] seems to be referring to [7].

hp-early-branch

nit: The output for git log --oneline --graph --decorate --all differs in manual setup and gitmastery setup. This is likely caused by commit ordering.

@damithc Is this a concern? I don't think it affects the hands-on. This is also present in other hands-on.

T6L2

hp-merge-ff

Suggestion: we can introduce the use of the flag --ff and --ff-only, as there is no place in T6L2 that uses this flag, and students might not be aware of it.

T6L4

hp-branch-rename

The output generated from git log --oneline --decorate --graph --all differs slightly from the expected output. Though they are technically the same, it could still be better for students if the expected output is exactly what they see on their screens to avoid confusion.

This is the actual output generated for both manual and gitmastery setup.

* 8060db5 (fantasy-books) Add fantasy.txt
| * 905b0ee (HEAD -> main) Merge branch textbooks
|/| 
| * 7513cea (study-books) Add textbooks.txt
|/  
* e6530ec Add horror.txt

T6L5

hp-branch-delete

Similar as above.

This is the actual output generated for both manual and gitmastery setup.

* 1114d75 (fantasy) Add fantasy.txt
| * 4af033d (HEAD -> main) Merge branch textbooks
|/| 
| * 70a8322 Add textbooks.txt
|/  
* 2b065b0 Add horror.txt

---

Will continue testing the exercises later on.

[damithc]

Thanks for testing this, @jovnc

T6L1

Misc

From the graph, it appears that the repo’s working directory reflects the code at commit c instead of b.

Good catch. The steps were a bit mixed up. Fixed.

hp-early-branch

nit: The output for git log --oneline --graph --decorate --all differs in manual setup and gitmastery setup. This is likely caused by commit ordering.

@damithc Is this a concern? I don't think it affects the hands-on. This is also present in other hands-on.

Yes, there is a discrepancy. Not sure why though, as both the script and the manual steps seem to be in the same order. Furthermore, the graph from the manual steps seems to be the correct one (the Add Martina to tennis.txt commit should appear at the top, as it is the latest commit). Any idea why the script produces a different result?

[jovnc]

@damithc This is likely caused by git log using the timestamp of the commits under the hood to determine ordering. It is likely because we generated the commits so fast, the timestamp is identical, those commits have a “tie” on the primary sort key.

I don't think there's an easy fix for this, unless we want to use time.sleep to guarantee ordering but it makes the download script even slower, which isn't ideal.

[damithc]

@damithc This is likely caused by git log using the timestamp of the commits under the hood to determine ordering. It is likely because we generated the commits so fast, the timestamp is identical, those commits have a “tie” on the primary sort key.

@jovnc possibly. Perhaps you can test this theory by inserting a short delay between commits? If it is indeed the case, we can insert a sleep only in places where it matters, such as this case.

[damithc]

T6L4

hp-branch-rename

The output generated from git log --oneline --decorate --graph --all differs slightly from the expected output. Though they are technically the same, it could still be better for students if the expected output is exactly what they see on their screens to avoid confusion.

T6L5

hp-branch-delete

Similar as above.

Both are caused by commits having identical time stamps. I've added a seep 1 in the manual setup to prevent this issue. Let's add a similar sleep in the HP setups.
Cases identified so far: hp-branch-rename, hp-early-branch, hp-branch-delete

[jovnc]

@damithc I have opened a PR to fix the known hands-on for now, verified that it is now the expected git log revision graph.

#213

[damithc]

hp-merge-ff

Suggestion: we can introduce the use of the flag --ff and --ff-only, as there is no place in T6L2 that uses this flag, and students might not be aware of it.

@jovnc Thanks for the suggestion. Done.

[jovnc]

Test Summary (Exercise)

T6L1

side-track

• @damithc Should be "(not master)" instead

• Verify logic can be improved, after completing step 1, the error message I get is the following, which should be a check at step 4, assuming students follow the steps. Ideally, we can refactor this such that the verification logic follows the step for a more logical flow.

@damithc I can open an issue for this, this seems like a good first issue for CS3281 contributors since I would say its severity is low and shouldn't be that difficult. What do you think?

branch-previous

• @damithc Should be "(i.e., commit Describe location)"

T6L2

branch-bender

• Some issues with verify logic:

1. git merge feature/login (fast forward by accident)
2. git reflog
3. git reset --hard (to undo the fast forward)
4. git merge feature/login --no-ff
5. git merge feature/dashboard
6. git merge feature/payments

Expected: success
Actual:

branch-forward

• After downloading, it is initialised with with-ginny branch, should it be initialised with main instead?

• Verify logic is wrong, actual repository also has the additional commit Set Initial State which is not accounted for in verify logic, which causes it to fail.

T6L3

✅

T6L4

✅

T6L5

✅

[damithc]

side-track

• @damithc Should be "(not master)" instead

@jovnc Fixed.

branch-previous

• @damithc Should be "(i.e., commit Describe location)"

Fixed.

@damithc I can open an issue for this, this seems like a good first issue for CS3281 contributors since I would say its severity is low and shouldn't be that difficult. What do you think?

Yup, that can work 👍