| ID: | 4787 |
| From: | rn...@igor2.repo.hu |
| Date: | Wed, 31 Mar 2021 18:43:50 +0200 (CEST) |
| Subject: | Re: [pcb-rnd] repo.hu server upgrade |
| in-reply-to: | 4785 from ka...@aspodata.se |
| replies: | 4789 from ka...@aspodata.se |
On Wed, 31 Mar 2021, karl@aspodata.se wrote: >I have subversion repo checked out here. >Igor have made a pcb-rnd/trunk/util/bisecter/bisecter. >Can that be used to bisect svn so I can see what revision broke this ? >Unfortenately, subversion takes some time to compile. >Is there some way to compile only whats needed ? That script has two parts: one is just an interactive algo that tells you what revision you should try, then you tell it what revision you really tried and whether it was good or bad. So it does the paperwork of keeping track on the current shortest path in besection. This part should work on any repo and any VCS system with any sort of linearity. The other part is pcb-rnd specific - it tries to make sure you have the right externs at the right version and caches the compiled pcb-rnd executable so you don't need to recompile the same version for a different bisect attempt. This part won't work with anything else but pcb-rnd. What I think we should do is this: 1. run the svn remote dump manually, to get the whole thing in a file 2. run the svnadmin import manually on this file, to see if it can import. There are two outcomes possible: A. it breaks; then what we have here is a "too high" version of svn rdump client can't properly communicate with a "too low" version of svn server. If we see that, we should probably ask the svn devs, or maybe report it as a bug - dump should be a format that works across versions, especially when emitted by old and imported by new. B. it doesn't break; in that case we have a flow control problem on the pipe between the two processes. Maybe too much stuff is coming in from the net and svnadmin ic consuming it too slow and the tcp connection breaks, oslt. If we can prove that, we will nee to switch to svnsync. Best regards, Igor2
Reply subtree:
4787 Re: [pcb-rnd] repo.hu server upgrade from rn...@igor2.repo.hu
4789 Re: [pcb-rnd] repo.hu server upgrade from ka...@aspodata.se
4790 Re: [pcb-rnd] repo.hu server upgrade from Barath Aron <ba...@mailbox.org>