Bug#911185: commit hash not found in translation-check header breaks the wml build
On Fri, Oct 19, 2018 at 03:06:43PM +0200, Laura Arjona Reina wrote: > I have done some tests to include some parts of what is done via cron scripts > to be done in .gitlab-ci.yml, to minimise the areas of webwml failing in the > CI, but later I thought that if those processes are indeed in separated cron > jobs that only run 6 times a day (and some others, once a day), maybe it's > not reasonable to make them run in salsa with each commit... Point. Perhaps another strategy might be better: For each file that is not committed to the repository, create a "sample" file that contains examples of what the data is supposed to look like, but that doesn't necessarily reproduce all the data. Then, create a Makefile rule along the following lines: Mirrors.masterlist: Mirrors.masterlist.sample if [ ! -f $@ ]; then cp $< $@; else touch $@; fi This copies the "sample" file to the "normal" file, but not if the normal file already exists. That way, the bogus sample files should never overwrite the real data, but at least CI will have processed some information (I think it's perfectly fine if a CI preview contains some bogus data). On www-master, the external cronjobs would remain, so there the website would contain the real data. -- To the thief who stole my anti-depressants: I hope you're happy -- seen somewhere on the Internet on a photo of a billboard
Bug#911185: commit hash not found in translation-check header breaks the wml build
Hello El 19/10/18 a las 12:19, Wouter Verhelst escribió: > On Thu, Oct 18, 2018 at 12:27:35PM +0200, Wouter Verhelst wrote: >> On Tue, Oct 16, 2018 at 11:52:55PM +0200, Laura Arjona Reina wrote: >>> * I'm not sure if we want that, or we prefer that the build fails and >>> make the CI not ignore these errors (then I guess the person committing >>> the file with wrong translation hash would get a mail, and hopefully fix >>> the file. If the page is built, maybe these kind of issues pass by >>> unnoticed). >> >> Yes. I think that should be done, indeed, and I've been working on it. > > FWIW, that's !33. I've been adding a "weak error checks" mode, where the > intention is that CI will fail, unless something is expected not to be > in the repository, such as the "Mirrors.masterlist" file, which seems to > be used all over the repository. > Mirrors.masterlist file I think it's the easier one to get, it's in the repo https://salsa.debian.org/mirror-team/masterlist.git/ so we can get it cloning the repo and putting the file in /english/mirror folder. > One issue I'm running against now is /english/international/l10n/data; > that directory seems almost empty, but it would appear that it is > supposed to contain configuration data for the website itself (i.e., not > data that I would expect to find in another repository). > > Why is that not just added to git and committed? > The files are generated by cron scripts run daily: https://salsa.debian.org/webmaster-team/cron/blob/master/lessoften-parts/1l10n-data I have done some tests to include some parts of what is done via cron scripts to be done in .gitlab-ci.yml, to minimise the areas of webwml failing in the CI, but later I thought that if those processes are indeed in separated cron jobs that only run 6 times a day (and some others, once a day), maybe it's not reasonable to make them run in salsa with each commit... Kind regards -- Laura Arjona Reina https://wiki.debian.org/LauraArjona -- Laura Arjona Reina https://wiki.debian.org/LauraArjona Sent with K-9 mail
Bug#911185: commit hash not found in translation-check header breaks the wml build
On Thu, Oct 18, 2018 at 12:27:35PM +0200, Wouter Verhelst wrote: > On Tue, Oct 16, 2018 at 11:52:55PM +0200, Laura Arjona Reina wrote: > > * I'm not sure if we want that, or we prefer that the build fails and > > make the CI not ignore these errors (then I guess the person committing > > the file with wrong translation hash would get a mail, and hopefully fix > > the file. If the page is built, maybe these kind of issues pass by > > unnoticed). > > Yes. I think that should be done, indeed, and I've been working on it. FWIW, that's !33. I've been adding a "weak error checks" mode, where the intention is that CI will fail, unless something is expected not to be in the repository, such as the "Mirrors.masterlist" file, which seems to be used all over the repository. One issue I'm running against now is /english/international/l10n/data; that directory seems almost empty, but it would appear that it is supposed to contain configuration data for the website itself (i.e., not data that I would expect to find in another repository). Why is that not just added to git and committed? -- To the thief who stole my anti-depressants: I hope you're happy -- seen somewhere on the Internet on a photo of a billboard
