tech-pkg archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: libncompat optimization for bootstrap (was: problems using an add-on compiler to bootstrap on FreeBSD)



At Thu, 9 Jan 2025 08:27:28 +0000, Jonathan Perkin <jperkin%pkgsrc.org@localhost> wrote:
Subject: Re: problems using an add-on compiler to bootstrap on FreeBSD
>
> * On 2025-01-08 at 21:47 GMT, Greg A. Woods wrote:
>
> > BTW, it is very dismaying how many times libnbcompat is configured and
> > built by the bootstrap, and, to "add insult to injury", without even any
> > configure cache!
>
> https://github.com/TritonDataCenter/pkgsrc/compare/e01dc9b4...feature/nbcompat/trunk

so, that looks pretty good, at least on first glance (I haven't actually
tried it)

Any reason it can't be pushed to the pkgsrc trunk sooner than later?


As an aside, I just went on a wild git-goose-chase trying to look at
that branch directly in a git clone.  Or alternately, how I came to love
gitk again!

I had assumed your branch naming convention was to put the starting
branch name after the feature name, i.e. here it should be "trunk".

However everything I tried seemed to suggest that "e01dc9b4" was not
actually on the trunk branch, and after cursing git for hours about how
it doesn't have easy tools to look at the history of a branch in
isolation (unless you know the starting branch or commit -- the latter
of which is in the github URL you gave of course, but I wanted to know
and use the branch name too).

I tried looking at the graph from "git log --all --oneline --graph
--decorate=full", but following the vertical lines in such a long list
in a pager was effectively impossible; and trying to load the entire log
into emacs magit was also effectively impossible as well; but good old
gitk can read all nearly 2 million log entries, and all the details
about them, in a "reasonable" time (on a sufficiently provisioned
machine), and then one can navigate around quite easily, and see the
branch membership of any commit with ease.  The resulting wish process
is pretty big, at over 9GB, but it has used under 6 minutes CPU time
even with me playing around in it (this is on Xen VM running with 8 3GHz
Xeon vCPUs and with 22GB RAM provisioned).

So it looks like the most likely starting branch was release/2022Q4, or
maybe joyent/release/macos.  I guess originally it might have been
branched from trunk, but some merging reshaped the DAG....

--
					Greg A. Woods <gwoods%acm.org@localhost>

Kelowna, BC     +1 250 762-7675           RoboHack <woods%robohack.ca@localhost>
Planix, Inc. <woods%planix.com@localhost>     Avoncote Farms <woods%avoncote.ca@localhost>

Attachment: pgpYZUDD6KINd.pgp
Description: OpenPGP Digital Signature



Home | Main Index | Thread Index | Old Index