Test for unresolved references when compiling the manual via makedoc.g
#84
Conversation
|
Sounds useful! |
|
Possibly works, since a package manual compiles only once, unlike GAP manuals which are compiled twice to get cross-refs right. What about packages which have more than one manual book though? E.g. GAPDoc or https://github.com/gap-packages/sonata - although it's not a suitable example, it uses |
|
This proposal would so far only apply to packages whose documentation can be compiled via the The main possible problem that I see with this change is that there may well be packages that currently do have unresolved references, (or people doing releases who don't necessarily have the right manuals compiled) but who don't care about that fact. This change would add a stumbling block for them. |
| if not IsPackageMarkedForLoading("$PKG", "") then | ||
| SetPackagePath("$PKG", "."); | ||
| fi; | ||
| Read("makedoc.g"); | ||
| GAPInput | ||
| ! grep -E "WARNING: non resolved reference" makedoc.log | ||
| rm -f makedoc.log |
There was a problem hiding this comment.
Shouldn't this also be done for the second run of makedoc.g? After all, they happen in different directories, and so it could happen that in one run the references can be resolved, but in the other they can't.
|
I didn't realise until now, this topic was already raised in issue #40. I'll think further about the best way of handling this. |
wilfwilson
force-pushed
the
doc-test-for-unresolved-references
branch
from
7f44c98
to
496e9c5
Compare
wilfwilson
force-pushed
the
doc-test-for-unresolved-references
branch
from
496e9c5
to
94e8835
Compare
wilfwilson
changed the title
makedoc.g
|
I will just limit this PR to dealing with the case that the manual is compiled via I want to add something to this PR before I consider it to be finished:
|
wilfwilson
force-pushed
the
doc-test-for-unresolved-references
branch
from
94e8835
to
ad4e02b
Compare
|
I think this should be alright now, but seeing as there is no automated testing for this repository, I want to test the script manually with this PR before the PR gets merged. |
|
Thank you for working on this. Yeah, lack of testing here is really bad. But I can't find the time and energy to set that up, sorry (perhaps one distant day during a GAP Day coding sprint... but soooo many other things seem more important... sigh). |
|
Thanks. I'm happy now. |
|
@fingolfin I don't have the permissions to merge this, so please could you do it. |
This is a proposal; I saw that @james-d-mitchell had added such a think to the Semigroups package's
.releasescript, and I thought it was a good idea.As a first step before investing any further work on this: do you think it's a good idea as a general feature? Could you imagine wanting to include this, possibly only enabled by an option?