tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Igal Sapir <>
Subject Re: Finally getting around to switching to Git
Date Thu, 25 Apr 2019 18:42:55 GMT

On 4/25/2019 11:30 AM, Coty Sutherland wrote:
> On Thu, Apr 25, 2019 at 2:06 PM Igal Sapir <> wrote:
>> On 4/25/2019 10:56 AM, Coty Sutherland wrote:
>>> On Thu, Apr 25, 2019 at 1:32 PM Christopher Schultz <
>>>> wrote:
>>>> Hash: SHA256
>>>> Igal,
>>>> On 4/23/19 12:52, Igal Sapir wrote:
>>>>> Another thing that I have changed in my workflow based on Mark's
>>>>> past suggestion, is that I keep a local repo for each major branch
>>>>> now.
>>>> Okay, I have done the following:
>>>> 1. Fork tomcat master to my own GitHub account
>>>> 2. git clone URL
>>>> 3. edit/add/commit/push
>>>> 4. Create a PR
>>>> I'm sure I can import the PR into tomcat-master. No problem.
>>>> Now, when attempting to keep my fork current, I've always done
>>>> something like:
>>>> git remote add upstream master-url
>>>> git checkout master
>>>> git fetch upstream
>>>> And I'm all up-to-date.
>>>> When I did that, I ended up bringing-down the 7.0.x and 8.5.x branches
>>>> as well. How can I limit the upstream to just the master?
>>> You can set the branch for your remote to master (or do it when you
>> clone)
>>> which should ignore other branches:
>>> git remote set-branches upstream master
>>> Then optionally configure --no-tags in your git config (or use --no-tags
>>> each time you git-fetch):
>>> git config --add remote.upstream.tagOpt --no-tags
>>> Then try fetching to verify it worked:
>>> git fetch upstream [--dry-run]
>>>> Or does my fork have to have everything, but I have to checkout a
>>>> single branch? If so, I'm not sure how to do that.
>>> It doesn't, but by default a `git fetch` pulls down all new work that
>>> exists on the remote, but not your local clone.
>> I am sure that Coty knows git better than I do, so if he says that it
>> doesn't then I stand corrected.
> I don't know about that :) If you do a regular `git clone apache/tomcat` it
> will pull the master branch and then references/histories for all remote
> branches which for tomcat is about a 100M .git directory. If you clone a
> single branch with no references such as `git clone apache/tomcat -b master
> --single-branch` then you get just the references/history for the master
> branch which results in about a 70M .git directory.

So one needs to consider whether that added layer of complexity is worth 
the savings of 30M disk space, even when multiplied over 3 branches.  
Imagine the reclaimed space after deleting the local SVN directories ;)

I think that working with the as the 
origin will make things much easier for Chris.  Especially with keeping 
the local repo up to date with the origin since he wouldn't need to use 
his fork.



To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message