When we introduced RedMine, Github issues were very limited. Calm
down, this is just a review of what's missing from Github issues, I am
not planning to propose or do anything :-) Let me do quick review of
missing features as of today (fall 2017) and possible mapping to
issues:
REDMINE - ISSUE
Status = Label
Priority = Label
Assigned To = Assignee
Category = Label
Target version = Milestone
Difficulty = Label
Found in release = Label
Votes = Reaction (+1)
Related issue = Issue link
Issue type = Subject (e.g. [TRACKER])
Bugzilla link = ???
So theoretically, we could map mostly everything to labels. There are
two ways of creating those labels, free form or with prefix. The
latter would look like:
That looks ugly, frankly I'd prefer free form, because we often only
set just few (if not none) of the flags - Found In Version is one of
the most important ones and that could be as simple as "In 1.16".
So basically, we are missing bugzilla link and private comments. Is
the BZ link item only informative, or is there some non-human
processing behind? I know there is a bot but does it need BZ link? It
could search comments for some token for the same thing. For private
issues we could use Bugzilla (usually security bugs only), there is no
solution other than pay them for private repo.
Or maybe I miss the main reason why we are not using Github issues at all?
I like github integration with PRs, speed and good reliability (only
few blackouts per year) and also new features like projects. On the
other hand, it's full commitment to something not under our control
(today we can easily move our git somewhere else, but we still loose
all PRs).
This email is just to discuss possibilities, I know that migration to
Github would be painful and even too expensive or perhaps technically
not doable (how to migrate so many tickets). It's a pitty that github
is now getting features it really needed.
I also really like gitlab which is packed with super nice features,
theoretically migration to something like that would be easier (open
source). On the other hand, we'd need to host this and one thing is
having redmine down for an hour, different thing is inability to push.
But this is definitely a possibility, we also have some know how
already running our internal instance.
Or maybe I miss the main reason why we are not using Github issues at
all?
As you said, the last time we evaluated it, it simply wasn't suitable.
The situation is more comparable now, however (in addition to BZ link
and private issues) I think we would also lose flexibility. In the the
last week or so, Marek and Walden have both proposed new plugins that
could be added to our Redmine - not possible if we go to GH, we'd be
stuck with waiting for them to implement new features (and in my
experience that's not fast at all). That's always the downside of going
proprietary over open source ;)
I like github integration with PRs, speed and good reliability (only
few blackouts per year) and also new features like projects. On the
other hand, it's full commitment to something not under our control
(today we can easily move our git somewhere else, but we still loose
all PRs).
Side note: Actually the PR data is accessible over the API, and I have
*all* of it in a MySQL DB. Yes, that's a lot of data - once I learn more
about data analysis (studying R at the moment :P) I will be doing things
with it.
This email is just to discuss possibilities, I know that migration
to Github would be painful and even too expensive or perhaps
technically not doable (how to migrate so many tickets). It's a pitty
that github is now getting features it really needed.
I think that's the key point. There's no doubt we *could* make GH Issues
fit our workflow (or any other bugtracker) - but the effort to migrate
20,000+ Redmine issues to multiple repos, as well as change all the
automation, is more than likely not worth it. There needs to be a *huge*
win for moving to GH Issues to make it happen, and I'm only seeing
side-grades and incremental stuff, I'm afraid.
I also really like gitlab which is packed with super nice features,
theoretically migration to something like that would be easier (open
source). On the other hand, we'd need to host this and one thing is
having redmine down for an hour, different thing is inability to
push. But this is definitely a possibility, we also have some know
how already running our internal instance.
I'd be +1 for GitLab-for-everything, assuming we can figure out some
reliablity, as you say. Maybe using Gitlab.com is an option. But *wow*
that is a lot of work :D
Dne středa 29. listopadu 2017 11:09:03 CET, Greg Sutcliffe napsal(a):
···
On 29/11/17 09:33, Lukas Zapletal wrote:
Or maybe I miss the main reason why we are not using Github issues at
all?
As you said, the last time we evaluated it, it simply wasn't suitable.
The situation is more comparable now, however (in addition to BZ link
and private issues) I think we would also lose flexibility. In the the
last week or so, Marek and Walden have both proposed new plugins that
could be added to our Redmine - not possible if we go to GH, we'd be
stuck with waiting for them to implement new features (and in my
experience that's not fast at all). That's always the downside of going
proprietary over open source ;)
I like github integration with PRs, speed and good reliability (only
few blackouts per year) and also new features like projects. On the
other hand, it's full commitment to something not under our control
(today we can easily move our git somewhere else, but we still loose
all PRs).
Side note: Actually the PR data is accessible over the API, and I have
*all* of it in a MySQL DB. Yes, that's a lot of data - once I learn more
about data analysis (studying R at the moment :P) I will be doing things
with it.
This email is just to discuss possibilities, I know that migration
to Github would be painful and even too expensive or perhaps
technically not doable (how to migrate so many tickets). It's a pitty
that github is now getting features it really needed.
I think that's the key point. There's no doubt we *could* make GH Issues
fit our workflow (or any other bugtracker) - but the effort to migrate
20,000+ Redmine issues to multiple repos, as well as change all the
automation, is more than likely not worth it. There needs to be a *huge*
win for moving to GH Issues to make it happen, and I'm only seeing
side-grades and incremental stuff, I'm afraid.
I also really like gitlab which is packed with super nice features,
theoretically migration to something like that would be easier (open
source). On the other hand, we'd need to host this and one thing is
having redmine down for an hour, different thing is inability to
push. But this is definitely a possibility, we also have some know
how already running our internal instance.
I'd be +1 for GitLab-for-everything, assuming we can figure out some
reliablity, as you say. Maybe using Gitlab.com is an option. But *wow*
that is a lot of work :D
Greg
+1 to Greg reply, personally I'm fine with redmine.
