fded6ce46cb167faaae559ff93b050c2b7d18ff1 max Mon Jun 26 08:59:00 2023 -0700 Porting hgGeneGraph to python3. refs #31563 diff --git src/hg/pyLib/README.txt src/hg/pyLib/README.txt index 5373006..61335df 100644 --- src/hg/pyLib/README.txt +++ src/hg/pyLib/README.txt @@ -1,19 +1,15 @@ This directory contains Python libraries for genome browser CGIs -written in Python (e.g. hgGeneGraph). The ideal Python version is 2.7, but -CentOs is at the time of writing on 2.6 so hgLib.py is avoiding any special 2.7 -features and uses the python2 in the hash bang line, so it will use whatever -python2 version is available (2.7 on genome-asia/genome-euro and 2.6 on the RR, -at the time of writing). +written in Python (e.g. hgGeneGraph). They were ported to Python 3. -To make the library work on python3, it uses six, which is included, as it is not -part of the standard library and a short 900+ lines module and also under a public -domain license. +The directory contains pymysql, so browser installations do not have install this +module using pip. pymysql is in the public domain, see pymysql/LICENSE. The version +bundled here is 1.0.2. hgLib.py includes various pieces of code ported from kent source hg/lib, like the bottleneck client, cart parsing, mysql functions, etc. The CGIs are not currently using a more efficient way of running Python scripts through Apache (WSGI), because this would be a considerable amount of administration across our various servers. It seems that running CGIs the "old" way is costing us only 100-200msecs so it is still bearable for the non-essential CGIs we may have in Python.