Bitcoin Unlimited - Development discussion

awemany

Well-Known Member
Aug 19, 2015
1,387
5,054
So I am browsing through the latest commits on 0.11cf_stats and I see that you, @YarkoL, seem to have picked up writing some test cases?

I am wondering whether we want to rework commits to clean up the history (and make BU more palatable to @Gavin Andresen or whoelse?)
@theZerg: What's your plan here?
 
  • Like
Reactions: YarkoL and sickpig

YarkoL

Active Member
Dec 18, 2015
176
258
Tuusula
yarkol.github.io
@awemany ,
yep and more tests are waiting in the wings.
Perhaps you might like to contribute also, as the traffic shaping
code lacks tests too, and the last time I checked, XT - where
@theZerg originally committed that feature - did not have
them either.

So it would be great if we here at BU could help both ourselves and
XT to get rid of that "bad code smell" ( to quote @Gavin Andresen ;))
 
  • Like
Reactions: awemany

theZerg

Moderator
Staff member
Aug 28, 2015
1,012
2,327
Go for it WRT to tests, that would be great! I do not have any intention of hiding the commit history to make it more palatable to some. The history is the history and should not be changed.

I am aware that others have a different git commit philosophy than I do and if they become the Developer they are more than welcome to change how it is done.
 

sickpig

Active Member
Aug 28, 2015
926
2,541
something for us to argue about... something makes me think that this is configurable. Can you check?
my money on minrelaytxfee or mintxfee cmd line options
 

YarkoL

Active Member
Dec 18, 2015
176
258
Tuusula
yarkol.github.io
Maybe I was a little too understated in
bringing out the irony of a situation where both
XT and BU share a same set of untested code,
yet it is the latter that gets accused of smelling bad.
 
Last edited:

awemany

Well-Known Member
Aug 19, 2015
1,387
5,054
Go for it WRT to tests, that would be great! I do not have any intention of hiding the commit history to make it more palatable to some. The history is the history and should not be changed.
You said earlier you wanted to periodically rebase off of Core. That would be a history-rewrite in itself - so one could use it as an opportunity to clean up commits like the 'wip' one mentioned by @Gavin Andresen.

I am of course not asking for outright hiding commit history here, I rather think of calling the current HEAD the new 'dev' or 'tentative' branch and create a reworked history on master that will be BU release #2. The current release #1 could be accessible and well-referenced and explained as a dead-end stable branch forever or at least as long as people care.

I am aware that others have a different git commit philosophy than I do and if they become the Developer they are more than welcome to change how it is done.
Yes. Let me just say I respectfully disagree with keeping the worse warts in the history. I think of git history (- in a perfect world - !) as a well-described stack of changes that allows to figure out what goes on and informs things such as a git bisect session.
 
  • Like
Reactions: freetrader

chainstor

New Member
Aug 28, 2015
16
25
XT and BU share a same set of untested code,
yet it is the latter that gets accused of smelling bad.
All criticism should be taken in good faith, and used to make BU better.
 
  • Like
Reactions: YarkoL

freetrader

Moderator
Staff member
Dec 16, 2015
2,806
6,088
> Yes. Let me just say I respectfully disagree with keeping the worse warts in the history. I think of git history (- in a perfect world - !) as a well-described stack of changes that allows to figure out what goes on and informs things such as a git bisect session.

Exactly. Cleaning up the code history can & should be a safe janitor-like activity carried out alongside the more messy development - I don't think it would need to be done by the original developers if their habit of working or time pressure means that they don't aggregate their commits, although that would be nice... All it would cause is a bit of extra work and require more branching.

From working in other projects, I have really come to appreciate clean atomic commits of bugfixes/functionality from those used to producing them like that.

There were other developers who didn't work like that.
In the end it if I recognise it as a habit of certain committers, it reduces the amount of attention I pay to their individual commits, since it's not possible to review as much in one go.

The projects still worked out fine with mixed working styles, though :)
 
  • Like
Reactions: awemany