+ ./install.py greatly expanded to ./env.py, a python script consolidating all manners of environment configuration including installation and docker option generation
+ issues are now required to use ./env.py to survey the environment
+ installs to arch/manjaro now supported
~ python no longer asks to install, it creates a virtual environment
~ centos7 -> centos-7 (for consistency)
~ scripts reorganized: i need that top level as pristine as humanly possible
- removed csv2html. just use a viewer

Co-authored-by: Ahmed Nofal <ahmed.nofal@efabless.com>

* Add perl-switch to Dockerfile

- needed for simple_cts script

* Fix Dockerfile

Signed-off-by: Vitor Bandeira <vitor.vbandeira@gmail.com>

- SDC file is carried across the stages with two variables SAVE_SDC and CURRENT_SDC to avoid having conditional statements on whether a stage is enabled/disabled 
- Moved the `sta.tcl` script under the openroad script directory 
- PDK commit is updated to propagate the new changes made to the `DEFAULT_MAX_TRAN` value https://github.com/RTimothyEdwards/open_pdks/pull/164
- Add base.sdc file to the manual_macro_placement test
  - The general sdc file doesn't support multiple clock ports so a custom one is needed for this design
  - This previously passed through the flow because sdc carry-over was conditioned on clock tree synthesis which is disabled for this design

openroad_app -> 7f23e5f6e6b8d36e6a8aaed6ad074902ea27c4ce

This changes Makefile environment variables to be consistent with caravel user project based on feedback from @mattvenn.

Also, the fastest_test_set/extended_test_set productions have been fixed from a break in #627.

* IMAGE_NAME -> OPENLANE_IMAGE_NAME

* CURRENT_TAG -> OPENLANE_TAG

~ PDK build now runs once, result reused
~ Designs run separately
- Removed outdated variables
- Cleanup now should always occur

~ Fixes #620

Co-authored-by: Harald Pretl <harald.pretl@jku.at>

- #607 - Lack of legalization was causing routing issues with OpenROAD.

- #601 - There is a cryptic "maze ripup wrong" message that makes aes_cipher fail: see The-OpenROAD-Project/OpenROAD#1110

- CU of aes_cipher reduced to 40 (50 was causing issues with TR)

- Fixed or_issue

openroad_app -> f7a44b69256c0b2a1dce50362997c4091d7f6a06

Co-authored-by: donn <donn@users.noreply.github.com>

Detailed placement is already done at this point once in the routing optimization and again if needed in the diode insertion strategy..This if statement just does detailed placement another time which sometimes makes the devices overlap, breaking the flow.

Reading sdc file from post CTS stage. Clocks already propagated so commenting line 46.

Co-authored-by: Donn <me@donn.website>

Fixes for the multi power domain generation, custom IO placement, and diode insertion strategy 4 (#609)

- Drawing multiple power domains wasn't working properly because `write_powered_def.py` assumed that the default vdd and gnd nets always exist at index 0. This isn't always the case because the nets are ordered differently in the def file. This PR fixes the issue by enforcing the index of the default power and ground net (the default power domain is assumed to be the first net specified by `::env(VDD_NETS)` and `::env(GND_NETS)`). 
- Added two new variables to give more control on the macros connection to the top level power grid.  `FP_PDN_ENABLE_MACROS_GRID` will either enable or disable connecting internal macros to power. 
`FP_PDN_MACRO_HOOKS` gives the ability to specify explicit power connections for internal macros. It is useful when the design has multiple power domains. For example, 
set ::env(FP_PDN_MACRO_HOOKS) "\
	mprj vccd1 vssd1"
will connect macro with instance name `mprj` to the power domain of `vccd1` and `vssd1`. This depends on https://github.com/RTimothyEdwards/open_pdks/pull/161 
- Added @20Mhz's fix for the custom IO placement 
- Updated the `place_diodes.py` to remove the dependency on the `new_int` API which is no longer available in openroad.

Some variables were moved to the PDK. PDK has also been updated.

* `report.sh` was rewritten in python and folded into `report.py`
    * 'E404' reverted- sorry @kareefardi, a lot of code just expects a number.
    * `report.py` was pretty much rewritten, for that matter.
* wns/tns reporting fixed

---

Also significant fixes made with the help of @Manarabdelaty 
* s44 pdn fix: weird substitution issue that somehow only spuriously surfaces
* manual_macro_placement_test pdn fix: FP_PDN_CHECK_NODES set to 0 (Warnings treated as errors by OpenLane, LVS still passes so we are golden)

This reverts commit 46a00b8b.

openroad_app -> 1ea38a76108667badbe79c088472b6162468fcce

- escaped_env_var is not reused correctly so that all escaped characters are correctly escaped

* Rework Multithreading Variables

~ ROUTING_CORES now overridable by environment
- Deprecated TR parameter file removed

* Update default value for ROUTING_CORES

* [BOT] Update openroad_app

openroad_app -> fb8ae93b6c7a5eb0e6fac83360a8a48d76c41885

* Updated `import opendb` to `import odb`

* Update Open_PDKs

Co-authored-by: Abanoub22196 <78970532+Abanoub22196@users.noreply.github.com>
Co-authored-by: mabrains <74658853+mabrains@users.noreply.github.com>

This fixes an issue with pip where support for file: was abruptly removed.

This addresses a race condition when a PR is merged when the scheduled CI is running.

Essentially, an image could be pushed with the tag master following a merge, where it would later be rewritten by the CI when it's done with the same master tag.

This makes it so the CI only pushes to a proper named tag and latest.

1. Save state before generating the report. Generating the report reads a run's `config.tcl` which is changed after saving the state, so we should save the state first then generate the report.
2. In `report.sh`, fetching some values happens before generating the values themselve. This adjusts the sequence of commands properly. 
3. Report `E404` instead of `-1` in values that fail to get captured as some of these values can be `-1` themselves.
4. Some indenting in `report.sh`

This adds a step to the CI that uploads log tarballs to the Google Cloud Platform based on a repository secret to help maintainers isolate issues that may only arise in a CI environment.

The relevant repository secret has already been added.

It can easily be extended to support AWS, Azure, etc as it leverages Apache libcloud.

The bucket being used is not public.

~ Reformatted run_designs.py to be more consistent with the rest of the codebase
~ Moved almost all top level scripts into other folders
~ Moved most top level documentation to docs/source
~ ol_install.py -> install.py
~ Fixed infinite recursion issue with or_issue.py