Обсуждение: enabling FOO=bar arguments to vcregress.pl
As I mentioned recently at <https://postgr.es/m/5d72f199-dc11-89a8-29d1-f20f9687c86f@dunslane.net>, I want to get USE_MODULE_DB working for vcregress.pl. I started out writing code to strip this from the command line or get it from the environment, but then it struck me that if would be better to implement a general Makefile-like mechanism for handling FOO=bar type arguments on the command line, along the lines of the attached. Thoughts? cheers andrew -- Andrew Dunstan EDB: https://www.enterprisedb.com
Вложения
On Mon, Nov 01, 2021 at 11:33:21AM -0400, Andrew Dunstan wrote: > As I mentioned recently at > <https://postgr.es/m/5d72f199-dc11-89a8-29d1-f20f9687c86f@dunslane.net>, > I want to get USE_MODULE_DB working for vcregress.pl. I started out > writing code to strip this from the command line or get it from the > environment, but then it struck me that if would be better to implement > a general Makefile-like mechanism for handling FOO=bar type arguments on > the command line, along the lines of the attached. I am not sure to understand how that will that work with USE_MODULE_DB which sets up the database names used by the regression tests. Each target's module has its own needs in terms of settings that can be used, meaning that you would still need some boilerplate to do the mapping between a variable and its command argument? -- Michael
Вложения
On 11/1/21 21:23, Michael Paquier wrote: > On Mon, Nov 01, 2021 at 11:33:21AM -0400, Andrew Dunstan wrote: >> As I mentioned recently at >> <https://postgr.es/m/5d72f199-dc11-89a8-29d1-f20f9687c86f@dunslane.net>, >> I want to get USE_MODULE_DB working for vcregress.pl. I started out >> writing code to strip this from the command line or get it from the >> environment, but then it struck me that if would be better to implement >> a general Makefile-like mechanism for handling FOO=bar type arguments on >> the command line, along the lines of the attached. > I am not sure to understand how that will that work with USE_MODULE_DB > which sets up the database names used by the regression tests. Each > target's module has its own needs in terms of settings that can be > used, meaning that you would still need some boilerplate to do the > mapping between a variable and its command argument? I think you misunderstood the purpose of my email. It wasn't meant to be complete patch. But here's an untested patch that should do almost all of what I want for USE_MODULE_DB. cheers andrew -- Andrew Dunstan EDB: https://www.enterprisedb.com
Вложения
On 11/2/21 11:03, Andrew Dunstan wrote: > On 11/1/21 21:23, Michael Paquier wrote: >> On Mon, Nov 01, 2021 at 11:33:21AM -0400, Andrew Dunstan wrote: >>> As I mentioned recently at >>> <https://postgr.es/m/5d72f199-dc11-89a8-29d1-f20f9687c86f@dunslane.net>, >>> I want to get USE_MODULE_DB working for vcregress.pl. I started out >>> writing code to strip this from the command line or get it from the >>> environment, but then it struck me that if would be better to implement >>> a general Makefile-like mechanism for handling FOO=bar type arguments on >>> the command line, along the lines of the attached. >> I am not sure to understand how that will that work with USE_MODULE_DB >> which sets up the database names used by the regression tests. Each >> target's module has its own needs in terms of settings that can be >> used, meaning that you would still need some boilerplate to do the >> mapping between a variable and its command argument? > > > I think you misunderstood the purpose of my email. It wasn't meant to > be complete patch. > > > But here's an untested patch that should do almost all of what I want > for USE_MODULE_DB. > > Now tested and doing what is expected. I'd like to backpatch it. Nobody who doesn't set USE_MODULE_DB would be at all affected, but I want to enable it for MSVC builds in the buildfarm. cheers andrew -- Andrew Dunstan EDB: https://www.enterprisedb.com
On Tue, Nov 02, 2021 at 05:59:49PM -0400, Andrew Dunstan wrote: > On 11/2/21 11:03, Andrew Dunstan wrote: >> I think you misunderstood the purpose of my email. It wasn't meant to >> be complete patch. >> >> >> But here's an untested patch that should do almost all of what I want >> for USE_MODULE_DB. > > Now tested and doing what is expected. I'd like to backpatch it. Nobody > who doesn't set USE_MODULE_DB would be at all affected, but I want to > enable it for MSVC builds in the buildfarm. Okay, now I get your intention here, thanks. Shouldn't you add some documentation to show out to use this facility when passing down parameters? -- Michael