[PATCH 3 of 4 V2] graphmod: use different styles for different edge types

[Durham Goode]

On 3/8/16 2:02 PM, Tim Delaney wrote:
> On 9 March 2016 at 04:42, Durham Goode <durham at fb.com 
> <mailto:durham at fb.com>> wrote:
>
>     Your change helps a bunch in the case where ancestors aren't
>     present, but that's pretty rare in our code base at least (since
>     we have very few merges).  Here is the before and after smartlog
>     output from my hg repo: https://bpaste.net/show/2093d7bad9c1
>     <https://urldefense.proofpoint.com/v2/url?u=https-3A__bpaste.net_show_2093d7bad9c1&d=CwMFaQ&c=5VD0RTtNlTh3ycd41b3MUw&r=nuarHzhP1wi1T9iURRCj1A&m=Ix7YLrzTSql7jLWOBs__ZDQLdJjoyuycoz3rbHzG16M&s=XWseVzXeEz1WRMYHFNEkVeXWNQ8rm7PMkOKRGdf50yg&e=>
>
>     I actually find the dashed line detracts from the general
>     glance-ability of the data, and am not sure we'd want to replace
>     smartlog with this output just yet.  Is there a config option to
>     disable these line stylings?
>
>
> How does it look to you if you replace colon (:) with exclamation mark 
> (!)? Note: need to be a little careful with find/replace to avoid 
> changing colons in the text. Personally, I don't like it, but it does 
> give a more solid but broken line.
Martijn tried it with a exclamation mark before (I saw an early 
prototype) and I think that was also not great (it was too subtle in 
some cases, and didn't really produce the dashed effect that would 
convey the not-my-parent notion).
>
> Another idea - do we need the entire path showing the grandparent 
> relationship? Would it be enough to just have it at the start (and 
> probably end) of the path?
This is what smartlog does, to some degree.  But smartlog inserts a full 
node (with no text) into the graph algorithm and draws an ellipsis on 
the line for that node.  Having ':' in a couple spots on the line might 
be interesting.  I'd probably not have the ':' as the first character in 
the line. Replacing the | on the blank line following the commit makes 
it obvious it's a break in the line that occurs just after the commit  
(having the : on the same line as some of the commit text drew my eye to 
that line and made me think the : was highlighting some aspect of that line.

@  4932aa  mjpieters
|  graphmod: end edges to missing parents early.
:
| o  d01268  durham
| |  parsers: optimize filtered headrevs logic
| :
| | o  55f8ae  durham
| |/   branchmap: check node against changelog instead of repo
| :
| |
o |  5cef34  durham  hgvip/release-2016-02-17
| |  test: remove dependency on $PYTHONDIR
| |
o |  4f99a5  durham
| | test: quick fix for test-hook
| |
o |  95bf01  quark  fbcrew/@
| |  chg: forward umask from client to server
| |
o |  e48d77  quark
| |  chgserver: add setumask method
| |

and

o    changeset:   32:d06dffa21a31
|\   parent:      27:886ed638191b
| |  parent:      31:621d83e11f67
| |  user:        test
| |  date:        Thu Jan 01 00:00:32 1970 +0000
| |  summary:     (32) expand
| :
o |  changeset:   31:621d83e11f67
|\|  parent:      21:d42a756af44d
| |  parent:      30:6e11cd4b648f
| |  date:        Thu Jan 01 00:00:31 1970 +0000
| |  summary:     (31) expand
| |
o |    changeset:   30:6e11cd4b648f
|\ \   parent:      28:44ecd0b9ae99
| ~ |  parent:      29:cd9bb2be7593
|   |  user:        test
|   |  date:        Thu Jan 01 00:00:30 1970 +0000
|   |  summary:     (30) expand
|  /
o |    changeset:   28:44ecd0b9ae99

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mercurial-scm.org/pipermail/mercurial-devel/attachments/20160308/4037ec8d/attachment.html>