<!DOCTYPE html>
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
</head>
<body>
<p>The danger of such a model is that branches quickly become stale
and a nightmare to rebase. In my experience this model only really
works well if developers are well synchronized, so people work on
largely distinct pieces of code, or everybody integrates their
changes with the trunk frequently, e.g. once per day.</p>
<p>The former doesn't really apply in this environment. The latter
one can already do when appropriate, since we have a trunk-based
model.</p>
<p>Note that in hg, <i>branches</i> mean divergent lines of work
that can't be re-based. What you describe here are <i>topics</i>
in hg.<br>
</p>
<p><br>
</p>
<p>Fabian<br>
</p>
<div class="moz-cite-prefix">On 5/16/25 12:34, Lawrence Paulson via
isabelle-dev wrote:<br>
</div>
<blockquote type="cite"
cite="mid:98F13930-4A94-4076-B447-C59B8CCF6736@cam.ac.uk">
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
<div class="BodyFragment"><font size="2"><span
style="font-size:11pt;">
<div class="PlainText">One thing I've noticed at Amazon is
that they insist on a single threaded revision trail. They
do this by requiring everybody to make their changes
within a personal branch, which they then re-base to the
main branch before pushing. Sometimes they also merge
adjacent commits. In return for mastering a couple of
extra commands, everybody benefits from a much cleaner
revision history. Do we want to try this?<br>
<br>
Larry<br>
</div>
</span></font></div>
<div><img src="cid:part1.brQ46jJf.S5q8Q5Sx@in.tum.de" class="">
</div>
</blockquote>
</body>
</html>