Opened 13 years ago
Closed 13 years ago
#1532 closed enhancement (fixed)
Have regress install the comments and throw an error if failure
| Reported by: | robe | Owned by: | strk |
|---|---|---|---|
| Priority: | high | Milestone: | PostGIS 2.0.0 |
| Component: | build | Version: | master |
| Keywords: | Cc: |
Description
I think if make check detects xsltproc, it should build and install the documentation comments.
The benefit of this check is that it catches bugs where we accidentally remove a function that we have documented or we change the signature of a function we have documented without updating the documentation.
Also right now, since the postgis and postgis_topology extensions package in the postgresql help, they will fail installation if the documentation and the function signatures do not agree. I think its a good thing as it insures our documentation is always consistent with what we have coded.
Change History (5)
comment:1 by , 13 years ago
comment:2 by , 13 years ago
Similar to #1529, if regress depends on comments, then regress also has to depend on XLSTPROC or things get even more complex and conditional in the build. If you're prepared for that, then we can give it a go.
comment:3 by , 13 years ago
| Priority: | medium → high |
|---|
I was saying that if XSLTPROC is installed then regress should test the comments otherwise it shouldn't. We've got a lot of conditional tests testing for those various versions of geos. Is it hard to make one that runs only if XSLTPROC is installed?
To be truthful, I don't care about this test, as long as you try to install postgis , postgis_topology using the extension module before you release an alpha tar ball. In fact I see no reason why the tar balls can't include already generated extension files since those files are the same regardless of if you are on Unix, Linux, MacOSX or Windows. I just don't want to come to the point of building a windows binary to distribute and "Yikes, gosh, darn it - the extensions won't install. I'll have to hack them or forgo distributing the binaries".
We've got lots of windows testers out there who rely on pre-made binaries and can help us squeeze out what bugs we have remaining. I want to USE THEM. :)
comment:4 by , 13 years ago
It looks like comments are not installed either, is there another ticket for that ?
comment:5 by , 13 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Done with r9164 (the ticket subject, not the "make install" issue)
Similar to #1549, if regress depends on comments, then regress also has to depend on XLSTPROC or things get even more complex and conditional in the build. If you're prepared for that, then we can give it a go.