| 1 |
On 2017-04-19 17:39, Gokturk Yuksek wrote: |
| 2 |
|
| 3 |
>> Display-If-Installed: app-backup/burp |
| 4 |
> Wouldn't you wanna limit this to <2.0.54 ? Otherwise this will pop |
| 5 |
> up for the consumers of 2.0.54 as well. |
| 6 |
Might as well, although at present there is no 2.0.54. |
| 7 |
|
| 8 |
>> /etc/burp/burp.conf . |
| 9 |
> You have an extra '.' at the end. |
| 10 |
Yes, it's a full stop at the end of the sentence. Come to think of it it |
| 11 |
looks confusing though, I'll remove it. |
| 12 |
|
| 13 |
> I think upstream using the new path is enough justification, do you |
| 14 |
> really need to justify it any further? |
| 15 |
It's not like that. Upstream has always used $sysconfdir/burp.conf , by |
| 16 |
default however that file provides client-mode configuration and bedup |
| 17 |
refuses to run unless it is pointed to a server-mode config file. Making |
| 18 |
bedup use $sysconfig/burp-server.conf by default is achieved in burp-1 |
| 19 |
ebuilds through the means of a Gentoo patch. |
| 20 |
|
| 21 |
> Maybe it's better to also provide a one-liner of 'mv' for people who |
| 22 |
> just want to upgrade to the new path. |
| 23 |
It is not the matter of the config file having been renamed, burp has |
| 24 |
always come with both burp.conf and burp-server.conf. |
| 25 |
|
| 26 |
> Overall, my impression is that people handle conf file changes in |
| 27 |
> pkg_postinst() with REPLACING_VERSIONS rather than news items. |
| 28 |
See above. |
| 29 |
|
| 30 |
> How fatal are the consequences of not updating the conf file path? |
| 31 |
> Would the program abort or misbehave? |
| 32 |
Abort. Please note however that in a typical production scenario bedup |
| 33 |
would be run periodically by cron or a similar tool, i.e. without admin |
| 34 |
intervention, meaning that for everyone who doesn't review their logs |
| 35 |
carefully backup deduplication would simply quietly stop working. |
| 36 |
|
| 37 |
-- |
| 38 |
MS |
Archives