On 03/27/2017 06:38 PM, Long Vu wrote:
Alternatively, just throwing an idea here, why Kallithea do not read
the config under /etc/mercurial/hgrc.d/ and other locations in same
order of precedence as the mercurial command line?
One reason is that the Mercurial config structure also to some extent is
used when hosting Git repos in Kallithea. It would be weird if Mercurial
config options kicked in when using Git ... and almost equally weird if they
only applied to Mercurial repos. That could be solved somehow.
Huh, why does Git read Mercurial config?? I don't know Git that much
but I would expect it also have its own config structure under /etc?
The "ui" table in the database contains entries that map directly to
Mercurial config entries. (It would make more sense if it was called
"config".) "ui" would be one reasonable value for ui_section for Mercurial.
Git itself does not read the "ui" table (and it is thus also out of the
question that it should read Mercurial global config) but "hooks"
entries are emulated for Git through a special Git hook (if I remember
correctly). The entry for 'paths' '/' is also used for Kallithea
globally ... so in that way the Mercurial-compatible settings is used
for Git repos too.
So ... there are some dirty details. But we could perhaps keep Kallithea
itself as before, while Mercurial could get additional configuration
when invoked directly from Python (and perhaps and perhaps not when the
hg command might be invoked from hooks).
I don't know which solution would be better:
* read all the same config files as Mercurial does and put db "ui"
values on top of that
* read a custom Kallithea Mercurial config file and put db "ui" values
on top of that
* put better UI on top of the "ui" table and just use that
/Mads
_______________________________________________
kallithea-general mailing list
[email protected]
https://lists.sfconservancy.org/mailman/listinfo/kallithea-general
