summaryrefslogtreecommitdiffstats
path: root/lib
Commit message (Collapse)AuthorAgeFilesLines
* ob_set_package_substvars(): DocumentPatrick McDermott2019-03-141-0/+10
|
* ob_get_binary_packages(): DocumentPatrick McDermott2019-03-141-0/+14
|
* ob_error(): Update @pure doc commandPatrick McDermott2019-03-141-1/+1
|
* ob_error(), ob_warn(), ob_info(): Make @stderr doc cmds more consistentPatrick McDermott2019-03-141-6/+5
|
* ob_get_*_parameter(): Fix @operand doc commandsPatrick McDermott2019-03-141-3/+3
|
* ob_init_package(): DocumentPatrick McDermott2019-03-141-0/+9
|
* ob_summarize_package_*(): RemovePatrick McDermott2019-03-131-14/+0
| | | | | | These were supposed to generate lists of architectures (platforms) and architecture (platform) wildcards for which a source package can be built, like the Architecture field of Debian dsc files.
* ob_get_doc_files(): DocumentPatrick McDermott2019-03-131-0/+8
|
* ob_get_doc_files(): Drop arch and plat parametersPatrick McDermott2019-03-132-8/+1
| | | | | | The set of documentation files to be provided should not be dependent on the architecture and platform for which at least some of the binary packages are built.
* ob_get_doc_package(): DocumentPatrick McDermott2019-03-131-0/+7
|
* ob_get_*_parameter(): DocumentPatrick McDermott2019-03-131-0/+15
|
* ob_parse_package_metadata(): Merge into ob_init_package()Patrick McDermott2019-03-131-15/+10
| | | | | | Now that caching is gone, the ob_parse_package_metadata() interface function is trivially short, and keeping it separate unnecessarily complicates the API.
* ob_get_binary_packages(): Use -p instead of -P for platformPatrick McDermott2019-03-131-2/+2
|
* ob_reduce_deps(): Use -p instead of -P for platformPatrick McDermott2019-03-131-3/+3
| | | | Same rationale as in commit 34f21a5.
* ob_parse_dep(): Swap -p and -P option lettersPatrick McDermott2019-03-131-11/+11
| | | | Be consistent with opkbuild and other executable CLIs.
* _ob_get_system_path(): Protect against DATADIR with "%"Patrick McDermott2019-03-131-4/+4
|
* Protect against cmd operands beginning with "-"Patrick McDermott2019-03-135-7/+7
|
* ob_parse_changelog(): Protect [ commands from strings beginning with "-"Patrick McDermott2019-03-131-4/+4
|
* ob_parse_package_metadata(): Return non-zero exit status on errorPatrick McDermott2019-03-131-1/+1
|
* Eliminate -a and -o options of [ commandsPatrick McDermott2019-03-132-11/+16
| | | | POSIX marks these as obsolescent.
* Protect [ commands from strings beginning with "-"Patrick McDermott2019-03-132-24/+24
|
* ob_parse_package_metadata(): Eliminate cachingPatrick McDermott2019-03-131-55/+0
| | | | Cf. commit da45b6d.
* ob_parse_changelog(), ob_parse_control(): Explicitly assign null stringsPatrick McDermott2019-03-132-2/+2
| | | | | POSIX say, "If value is not specified, the variable shall be given a null value." Make this explicit, for consistency.
* ob_substvars(): s/while true/while :/Patrick McDermott2019-03-131-1/+1
|
* ob_parse_dep(): Fix crash on dep without version, arches, or platsPatrick McDermott2019-03-131-4/+4
|
* Don't let libopkbuild errors cause exit with -ePatrick McDermott2019-03-134-26/+27
|
* ob_substvars(), ob_init_package(): Call _ob_error_msg()Patrick McDermott2019-03-132-2/+2
|
* ob_error(): Make non-fatalPatrick McDermott2019-03-131-3/+2
|
* libopkbuild: Abort on invalid function argumentsPatrick McDermott2019-03-139-353/+158
| | | | | | | | | | Shift arguments and abort instead of returning 125. Incorrect numbers of function arguments suggest application/library incompatibilities or serious errors by the application developer. Either way, the application developer should be made immediately aware of this (and not allowed to simply not check return values), and continuing to run and handle any further API calls may be unsafe.
* _ob_abort(): New internal function to crash hardPatrick McDermott2019-03-131-0/+7
|
* ob_match_arch(): Use clearer variable namesPatrick McDermott2019-03-131-9/+9
|
* ob_match_arch(): Replace generated ERE with native SCLPatrick McDermott2019-03-131-8/+20
|
* _ob_validate_*(): Replace echo with here-document in grep commandsPatrick McDermott2019-03-121-6/+11
|
* _ob_get_upstreamver(), _ob_get_distrev(): Replace echo with printfPatrick McDermott2019-03-121-2/+2
|
* _ob_get_doc_package_2(): Replace echo with printfPatrick McDermott2019-03-121-1/+1
|
* ob_reduce_deps(): Replace echo with printfPatrick McDermott2019-03-121-1/+1
|
* lib/package.sh: Update copyright yearsPatrick McDermott2019-03-121-1/+1
|
* ob_get_binary_packages(): Drop echo along with unnecessary assignmentPatrick McDermott2019-03-121-4/+2
|
* ob_get_binary_packages(), _ob_set_binary_parameter(): Replace echo and grep ↵Patrick McDermott2019-03-121-9/+10
| | | | with case
* ob_get_binary_packages(): Fix indentationPatrick McDermott2019-03-121-2/+2
|
* lib/control.sh, lib/package.sh: Replace echo with here-document in tr commandsPatrick McDermott2019-03-122-15/+41
|
* ob_parse_control(): Rewrite missing fields handlingPatrick McDermott2019-03-121-6/+7
| | | | Get rid of echo and sed.
* ob_parse_control(): Rewrite field checkingPatrick McDermott2019-03-121-23/+26
| | | | | | | Get rid of the echo and sed pipeline and use case constructs instead of "[" commands with "${parameter%word}" expansions. The new logic also will only report duplicate unknown fields once.
* *ob_[gs]et_*_parameter(): Make validation more immediately clearPatrick McDermott2019-03-121-20/+12
|
* _ob_set_*_parameter(): Don't expand escaped value early in evalPatrick McDermott2019-03-121-8/+2
|
* ob_get_*_parameter(): Drop echo and improve quotes in eval cmdsPatrick McDermott2019-03-121-2/+2
|
* ob_parse_version(): Improve eval safetyPatrick McDermott2019-03-121-4/+4
| | | | | | | There likely wasn't a vulnerability here since versions are validated first, but unnecessary expansions like this in eval commands are bad practice, and a bug in the validation could have led to a vulnerability here.
* ob_parse_dep(): Drop echo and improve quotes in eval cmdsPatrick McDermott2019-03-121-2/+2
|
* ob_substvars(): Replace echo in eval commandPatrick McDermott2019-03-121-1/+1
|
* ob_set_substvar(): Don't allow "_" in variable namesPatrick McDermott2019-03-121-2/+2
| | | | | The character was allowed due to the details of the implementation, but the SPF 2.0 specification doesn't allow it.