Fixes for building from git #8
Conversation
Version is now branch-tag-n-commit. branch is only included when it's not master. n and commit are only included when it's not the release version. n is the number of commits since, and commit is a short id of the current. -dirty gets appended in case there are uncommited changes.
readme.ptxt was missing. Too avoid maintaining 2 readmes, generate readme.txt from README.md
…and only evaluate it once.
| HG_ARCHIVE_FLAGS ?= -X .hgtags -X .hgignore -X .devzone -X scripts/make_changelog.sh | ||
|
|
||
| # Text processing and scripting | ||
| AWK ?= awk | ||
| GREP ?= grep | ||
| PYTHON ?= python | ||
| UNIX2DOS ?= $(shell which unix2dos) | ||
| UNIX2DOS_FLAGS ?= $(shell [ -n $(UNIX2DOS) ] && $(UNIX2DOS) -q --version 2>/dev/null && echo "-q" || echo "") | ||
| UNIX2DOS_FLAGS ?= $(shell [ -n $(UNIX2DOS) ] && $(UNIX2DOS) -q --version >/dev/null && echo "-q" || echo "") |
There was a problem hiding this comment.
I just ran into the problem that is fixed by this line (unix2dos producing stdout output, that ends up inside UNIX2DOS_FLAGS). Good to have it fixed, but I wonder if this fix should be in its own commit?
Regardless, this should probably redirect both stderr and stdout (so >/dev/null 2>/dev/null) to prevent issues when unix2dos decides to produce output on stderr too later?
There was a problem hiding this comment.
Made a new PR out of this: #11
|
I cherry-picked most of this PR apart over a few PRs, to make reviewing a bit easier. I want to seriously thank you for your efforts. This was really useful. The one thing I did not cherry-pick is the "markdown to txt" convertor. I do not think that is the right approach. But after merging of all the other PRs, that is something we should talk about in a better setting :) For example: a single PR that only adds that functionality. If you feel up for it, I would love to have a PR for it, to talk if we want to go that road. Tnx again! Going to close this PR for now (as I am pretty sure the other PRs will be merged :D) |
Source bundles do not yet work with git.
Note on why README.md is so diff: It's mostly UNIX/DOS linebreaks. I also based it on the previous readme.ptxt. It works with the current .md as well, but the markup will be slightly off. Also, there seemed to already be some minor issues with it.