Shape the future of FME with your ideas
New and Open ideas are awaiting review or have been reviewed by our Product Management team, and are open for commenting and voting.
Hi,I've just spend hours to track down an error in a custom Workbench transformer calling a Python transformer.Both transformers were fine, but I finally tracked the error down to being the name I'd given the Workbench transformer which contained a dash (as in "...T-SQL..."). FME used that name indiscriminately to auto-generate the code to call the Python transformer, yielding a not very clear error message, caused by dashes not being allowed in variable names in Python.FME do already replace blanks with underscores when naming its own transformers, so why not extend this to do the same with dashes (and all other illegal Python variable name characters) ?
I recently wanted to use the maximum value of an attribute on one table as the start value of the counter on another table. To do this i have to calculate the value then use an attribute creator on each table and a feature merger to then calculate the new counter valueI would have been much neater and simpler if i could have passed the value to a parameter that the counter could use as the start value.
As stated in the documentation of the ListBuilder, Features output from this transformer have no geometry.Hundreds of times already I have used the FeatureMerger to re-insert the geometry. During the #fmedays, @ArneBruksch informed me you can also use 'GeometryExctracter' and then 'GeometryReplacer'.Anyway, it would be nice to have the possibility to keep the geometry automatically while building a list. Any other fans?
In FME 2015, sending an invalid encoded geometry to the geometryreplacer causes the whole workbench to fail. It would be better to have a rejected port that handles these and lets the other geometries be produced.
***Note from Migration:*** Original Title was: True horizontal scaling based on load. FME Server Cluster to enable deployment of 10's, 100's or 1000's of engines for short duration.
Add a centre label position to the labeller transformer. Currently only left/right justification (and above below) justifications are available. Centre justification will make it easier to produce annotations for mapping data that stores annotations as lines (i.e. OS data in shapefile or DB formats).
Thumbnails can be embedded in HTML as explained in this link: https://knowledge.safe.com/articles/1176/embedding-images-in-html-with-fme.htmlHowever, it would be good to have this included in the HTMLReportGenerator. Maybe as part of the table or as a stand alone option. The images that are embedded are often not large enough to show detail and cannot be opened in a new browser by clicking on it.
Keep track of what paths data has been read from/saved to and make these easy to go back to. Every time a file open or save dialog would be in play, an easy pulldown of previously used paths would be available.
We are in an organisation with limited budgets and not as many licences as we would like. For the team to be able to learn FME and develop ideas they need access to the software. However, there isn't enough licences to go around, I'm sure this isn't unique to just our company. What I propose is a sandboxed online FME desktop that allows for people to learn and test ideas without a licence. You wont be able to send it to FME server or actually do any translations, all writers could go straight to a virtual data inspector for viewing purposes only. This way quite large organisations could teach up their staff without having to buy a lot of licences or downbloading trials. I realise this could be a lose of potential earnings for Safe and their re sellers, but it could also persuade more people to use FME and could be a great demonstration area. Also on the help forums and ideas section of the knowledge centre people could use these sandboxes as a way to demo their ideas and help people. I wander what the re sellers in the UK would think of this idea? @1spatialdave
Currently the NumericRasterizer seems to always work with 'Last Wins' mode. That is, the transformer uses z-coordinate of the finally coming vertex as each cell value, if two or more input vertices are within the same cell.It would be nice if the transformer could work with several modes such as Min, Max, Average, First Wins, Last Wins, Standard Deviation, etc. This might be useful especially to analyze a point cloud efficiently, in conjunction with some point cloud transformers.
I would love an FME Server jobs statistics transformer. Yes you can grab the engine.log and parse the data in and create stats ,e.g. average job times, but this is a little long winded. Either through the web portal ( with graphs and load data) or a dedicated transformer would be really useful for load testing FME server.
***Note from Migration:*** Original Title was: System Parameter to return OS platform on which the current FME session is running
Please could we have control over the name of the output table from an SQLExecutor.Currently it is set as Result and Initiator
To provide more functionality on working with lists I would like to propose a ListExpressionEvaluator transformer. This would evaluate a specific expression on every item in a list. The result can overwrite a list item or form a new attribute on the list.In the expression you would have access to the attributes of a list item, and all the attributes of the feature itself. As an example list:someList[0].Value1 = 15 someList[0].Value2 = 30 someList[1].Value1 = 15 someList[1].Value2 = 20The user would pick the list: somelist. The expression could be: MaxValue = max(Value1, Value2).Which would leave the list as:someList[0].Value1 = 30 someList[0].Value2 = 15 someList[0].MaxValue = 30 someList[1].Value1 = 15 someList[1].Value2 = 20 someList[1].MaxValue = 20Bonus points for providing access to previous/subsequent list items, similar to the adjacent feature access on the AttributeCreator. This can of course be accomplished with Python or a ListExploder, but preferably we won't need Python and I can keep my features intact.
It would be nice if there were a built-in transformer that would check if the supplied directory exists and is writeable and optionally create it if it doesn't exist. There are a few transformers that can write to files directly, and they will fail if the directory doesn't already exist. For example the HTTPCaller with Save Response Body to File.I do have a python script that does it, but wrapping it in a custom transformer get tricky with parameters [see python-caller-user-parametersl]
I recently read and tried to follow a tutorial by @DeWetAtSafe (link included below). It appears as though parts of the tutorial have changed in Amazon Web Services and I found myself wishing I could post a comment to get some help working through it given the changes. Link to tutorial - https://knowledge.safe.com/articles/24147/tutoria...Can you add the ability to comment on Knowledge Center articles?
I think it would be cool if transformer could have similar floating toolbars as readers and writers got in 2015 or simply another button next to the settings gear.Especially in transformers like the PythonCaller or SQLExecutor you often need to open the editor view (SQL Statements, Python Script) multiple times which requires several clicks.
Right now, Inspector only works if translation is successful.I suggest to make 'completed' Inspector show the results even the translation is Stopped or Aborted.
It would be great to show which readers or writers has change from previous versions. I just had a classic example of why it is needed. I have migrated a 2014 workbench to 2016. i tried running it and it failed on startup, with import arcpy, could not find arcgis. The user have no idea that the reader format as changed (in the case the Geodatabase reader has change significantly)After add the new reader format - it seems to worksame applied to the geodatabase writer
On the Writer it is possible to move a canvas FeatureType between writers using the 'move' function. With Readers though you have to reimport the FT onto a new Reader. It's very common that I'm given workspaces that have multiple unnecessary Readers defined, all with say 1 FT and for best practice its necessary to consolidate the Readers. Deleting and reimporting is quite a bit of effort so I wondered if an alternative could be explored. I appreciate this function would likely need restricting to being between 'like' formats.
Improve the Chopper transformer, so it becomes possible to chop a line at a specific coordinate / vertex
Rather than having to duplicate the Line (network) many times, once for each source point. Can it be built to handle more than one source. I have 1000 source points and a network of 100,000 lines. This blows out to 1,000,000,000 lines to process. Ouch!!! thanks Steve
Currently the copy number (_copynum) created by the Cloner always starts from 0. In some cases, the copy number will be used for parameters in the following transformers and there are cases it has to be a 1-based number. Although it's easy to increment the number, it might be better that the user could specify the starting number as a parameter of the Cloner, as same as the Counter.
Update or merge the various style transformers (PDF, Mapinfo, CAd etc) into a style manager super transformer. Allow for importing styles from other datasets and tables as well as from json and SLD (whole map styling) and SVG + PNG (individual feature styling) files. Various agencies such as the OS use sld style files. OS Mastermap for instance would be a lot less painful in organizing styling if it were possible to open the style file up and convert data in the fme-colour and other style fields. If this information auto transferred to other writing formats then even better. If you could also preview styles within the style manager such as point markers and transparency as well as open up new style markers in SVG or PNG format then FME would start to look as powerful as GIS in terms of map creation. It would be amazing if the style manager could connect to the Mapnik rasterizer as well.
FME Desktop / and or Server should use default python installation, so that all existing packages and or fmeobjects should be in the same python interpreter.I think the use C:WindowsSysWOW64python27.dll is not a very good implemetation (due to various installation of python).During installation of FME Desktop / Server it should be able to detect python installations (if none install a python interpreter), and then setup accordingly. so the custom / default python should be combinedWith this you then can choose between for example 2.7 32-bit and 3.4/3.5 64-bit etc.
In-depth courses for all FME skill levels.
Explore our extensive collection of how-to guides and tutorials.
Log an issue with us. Our support team is ready to help.
